$1K won in Q contest - ideas on how to trade it? Awhile back, I won$1K in the Q contest (probably just luck). I'd like to trade it (I know it is a total gamble at this level of capital):

• Suggestions for strategies?
• How I might use Q to research and develop a strategy?
• Trading platform options? E.g. Alpaca.markets? Robinhood? Unhedged.io (not live yet?), Etc. Obviously, I need free data and free trades with only $1K of capital. Any ideas are welcome. Thanks. 52 responses Hi Grant, Speaking as someone who is about to do exactly the same thing, I think the important things to pay attention to are: 1. High Position concentration (like 3-4 stocks max) 2. Low turnover 3. Long Only You're not going to manage to be as diversified as you would be able to do with$10M etc, so it's important the factor is very strong in the extremities. The stocks you are trading should also be highly volatile if you're looking for returns. The long only rule is a personal choice for me, but I think it's probably a valid rule if you're only wanting to risk the $1,000. With regards to live trading, I'm just going to be using the Q live trader and then manually transferring orders over (where the low turnover rule comes in) but I'm looking into other methods (probably Alpaca but I'm not sure yet). Unhedged.io looks interesting though, thanks for bringing that to my attention! I'm not sure if alphalens will be particularly helpful for this. Maybe it might be a good idea to focus on small/micro cap stocks and try to exploit big jumps maybe? That's where the returns are at least. Good luck with your trading :) Thanks Jamie - I guess so-called "penny stocks" (<$5 per share) would be one path. These should be in the Q database, and could be sliced and diced just like folks do for researching stocks for the Q fund.

I'm also wondering about the viability of ETFs, if anyone has ideas/experience.

By the way, I thought Q had done away with the "Q live trader" (live paper trading)? It should be equivalent just to run a noteboook/backtest periodically.

Not sure that the low turnover rule matters if there are no commissions (except that there are regulatory restrictions in how frequently one can trade e.g. pattern day trading rules). I think staying within regulatory bounds, one can trade weekly (or perhaps a bit more), which would be fine with me.

I'm trading through IB at the moment so commissions are important to me. With regards to the Quantopian Live Trader, it's still there but there's a 15 minute delay I believe? If turnover is low enough (10% ish a day it shouldn't be too much of an issue)

Penny stocks tend to be a lot less liquid though. Wider spread with less volume on each side, so you still have to account for slippage. The Q default slippage model may not model fills for illiquid stocks very well, so what might look good in a Q backtest may not work at all in the real market (market makers tend to own this space too).

Personally I would do long only as well, since I can’t leverage much and I think/hope I can stomach volatility and drawdown. Some sort of Value + Momentum strategy rebalancing monthly or quarterly even. Getting decent fundamental data might be difficult though. You’d most likely have to pay for it e.g from Quandl.

Good luck, and great to see you back in the forum Grant. When will we see you back in the contest again?

One option is to put into an income fund.

You can also write up the algorithm in quantconnect. They allow live trading of stocks, forex and futures.

@ Joakim -

Thanks for the reminder about slippage and fills. Presumably Q still supports a custom slippage model?

Regarding the contest and trying to get an allocation, my fear is that it'll be a huge time sink with low probability of reward. I've been communicating privately with Dan Whitnable on some ideas on how Q might move forward, which has been interesting. We'll see where it goes. I could comment further, but I'm restricted to just discussing technical information relevant to my interests on my threads, lest I create a distracting diversion that would reduce the number of viable algos produced by the crowd (basically consistent with the "Keep it quant finance" on https://www.quantopian.com/policies#code-of-conduct).

Fair enough. Yes, custom slippage is still available. I just think it would be very difficult to get it right. One might set it too aggressive for some illiquid stocks, and not aggressive enough for others. Also, liquidity might change from day to day, as well as intra-day. For highly liquid stocks this is not really a problem, unless you’re trading BIG size, but for penny stocks I think it would be. I could be wrong though.

Is there any way to analyze just ETFs on Q?

@Grant, you must be kidding, or pulling everyone's chain.

Can the following be said or should I delete it?

Trading on a $1k account is silly, especially with your talents. The pattern day-trading rule will limit your trades to 3 per week until you reach$25k.

Say you average out a 20% CAGR, it would take some 17.66 years to reach that level. Note that the average day-trader is in the 4.3% range, so I am generous here taking note of your skills. There is a consolation, after 21.46 years, you will have reached $50k, and after 37.88 years, your account would have grown to$1M. The buying power of $1M in 37.88 years will be much less than it is today. May I suggest you find a better use, not necessarily just of the$1k, but mostly, of your time. Use that $1k in the PR needed to outsource your best trading strategy which must support a much higher initial capital. Pattern Day Trading rule only applies to Margin accounts. I have an idea, wait a good opportunity to trade Vix, and with a little chance you can multiply by 5 your invest, much more if you place your stop near 9 ! I'm pointing out these two account types because it strikes me there are a remarkable number of big sites out there that are either confusing (overly complex) or vague about them. The two types of accounts and their restrictions, brief summary Cash: Reg+T rules, where T (time) might be 1 to 3 days, usually 2 nowadays. Have to wait that long to sell after buy. Margin: PDT rules. If the portfolio value is below$25k and there are more than 3 buy/sell of same stock within any rolling 5 day period, the account is flagged Pattern Day Trader and suspended. Above 25k, no problem.

Those are my current impressions, I believe it sums them up, and feel free to correct me.

They were enacted after the dot-com crash some 18 years ago with only about 600 comments on at least one of them.

Reg+T was of course useful back then because it took awhile for settlement as brokers sent trade notices to each other by--what--pony express or something apparently, before algorithms. Ok I don't know the details of their settlements but my hunch is these days our communications and systems make RegT unnecessary, and all it is serving to do now is keeping small investors out of the market, the opposite of Free Enterprise. The link by Jamie above also outlines arguments on both sides after the section he is pointing to, although with the two types jumbled together.

Starting small is not a bad idea in my opinion. If I were to trade my own account (no interest at the moment), I would ‘only’ be paper trading via the broker API (say IB for example) for quite a while at least 3-6 months. Only once I’m convinced that I have a strategy that is predictive on future data would I start with real money in the live market, and I would start super small, so 1k might be a good start. If, and only if, the strategy can be proven to be consistently profitable in the live market, using small amount of capital, would I throw more money at it.

In other words, starting with a large amount of capital from day 1, without enough proof that the model is consistently profitable on future data and in the real market, is likely to be disastrous, and in one word ‘silly’ in my opinion.

Good alpha research and a good backtest is a good start, but it’s far from any guarantee that the strategy will work on future data. And even if the strategy has been proven, with reasonable confidence, to work on future data, doesn't guarantee that it will actually be profitable in the real market, e.g. due to slippage.

Regarding ETF research, the filter in the attached NB should include all ETPs including all ETFs (would include leveraged ETFs as well, and possibly other stuff too).

If there are specific ETFs you want to research however, it might be best to define them in a security list (copied from the help doc).

def initialize(context):
# In this example, we're looking at 9 sector ETFs.
context.security_list = symbols('XLY',  # XLY Consumer Discretionary SPDR Fund
'XLF',  # XLF Financial SPDR Fund
'XLK',  # XLK Technology SPDR Fund
'XLE',  # XLE Energy SPDR Fund
'XLV',  # XLV Health Care SPRD Fund
'XLI',  # XLI Industrial SPDR Fund
'XLP',  # XLP Consumer Staples SPDR Fund
'XLB',  # XLB Materials SPDR Fund
'XLU')  # XLU Utilities SPRD Fund

0
Notebook previews are currently unavailable.

@ Guy Fleury -

@Grant, you must be kidding, or pulling everyone's chain.
Trading on a $1k account is silly, especially with your talents. I'm not kidding, but I do realize that there are some issues with trading relatively small amounts of capital. From the beginning, Q has been a curiosity and a hobby for me; it is no sillier than golf or building a ship in a bottle. If anything, Q and the community have come to take themselves too seriously. One rule I have for myself is that the hobby should be completely self-funded. Since I've only made$1K in capital so far, that's all I have to work with (I suspect if others interested in trading followed this rule, rather than speculating with money gained by actually working, i.e. producing value for others, the whole financial industry might collapse).

As a side note, last year, I went to the World Maker Faire in Queens, NY. The ethos is to make something from almost nothing just for fun. It's not about being practical (which most of us have in excess in our daily lives).

I suppose one solution to the regulatory problems would be to open a margin account with $25K, but lock up$24K in something prudent (e.g. an ETF that represents a chunk of my overall asset allocation).

There's also the issue of taxes. Trading in an IRA account might be the way to go, since paying short-term gains taxes will eat into profits, right?

I see that Alpaca.markets is working to offer a so-called "Power Account" which must be a margin account conforming to the regulations:

We are launching a new account type to unlock the unstoppable future intelligence. The Alpaca Power Account will provide you with the ability to short sell and leverage up to 4 times in staging.

It seems like if Alpaca offers this, and maintains zero-commission trading, it would be revolutionary (but maybe I'm missing something, or its just a teaser, and they'll end up charging commissions).

A note perhaps of interest to some:

https://docs.alpaca.markets/account-plans/

At this time, Alpaca Securities LLC only provides individual taxable brokerage accounts.

@Grant, say you have a trading strategy that is sustainable, marketable, and executable over a 30-year period. And since I know you have the talent to design something outstanding, I will give you a 30% CAGR. The math goes like this:

With $1k: 1,000 (1 + 0.30)^30 = 2,619,996 With$1M: 1,000,000 (1 + 0.30)^30 = 2,619,995,644

The difference should be considered your opportunity cost. The what you threw away simply because you could not get the extra capital to put your talent to work in the most efficient manner.

We all have to make choices, but here, I do think you undervalue yourself. You can do much better than starting with a $1k account since you will be wasting so much of your potential. @ Guy - For me, its an avocation (like fishing or woodworking or foosball). I have plenty of other serious things to be concerned with. We'll see...maybe I'll try the Q contest again, but it would also be from this perspective. Not much to add. Mostly agree with what has been posted. HFT will be fleecing you on each entry and exit unless you trade securities with tight spreads and are able to "hide in the crowd." So avoid strategies that have high turnover to make up for low per-trade profit margins, and avoid illiquid stocks. Don't bother going down the rabbit hole of trying to model HFT game theory within a Quantopian custom slippage model. You can't. As for strategy... I feel like with$1000 of free money on the line you may as well pursue return on capital over preservation of capital. Let the magic of compounding do its thing! YOLO

The advantage of having such a small account balance is that your trades won't have much market impact, so you don't need to spread yourself into weaker positions. Go all in on the positions where the signal is the strongest. You must have some strategies that have stood the test of time, no? How do they perform if you distill down to the 3-5 positions with strongest signals on the long side or something like that?

Obviously, be wary of overfits. And be wary of alpha that may appear to exist because the backtest doesn't properly take into account things like spreads, short availability, liquidity, execution quality, queue, fees, etc. Start slow and keep an eye out for incongruities between expected and actual performance.

I believe Alpaca accounts are "limited margin accounts" which means you only get 1x margin, but your trades settle instantaneously (instead of having to wait 3 days for the money to appear in your account after you sell a stock). This means PDT applies. Not an issue for a daily-rebalance strategy though.

@ Viridian Hawk - Thanks. Hadn't heard of imposter syndrome. Hope I don't have it, since I see that as an internalizing disorder, electroconvulsive therapy is a treatment option--no thanks; I'll pass.

Intuitively, I'd think that for individual small-time traders, thinly traded penny stocks would be profitable, since presumably institutional traders can't deal in them. They would be further away from the efficient market--more prone to irrational exuberance and FUD effects. But maybe this is wrongheaded, and how people lose their shirts (or in this case, a $1,000 shirt). @Grant -- I very quickly lost a$500 shirt by trading illiquid stocks while a version of the very same algo running on a Quantopian paper trade was simultaneously printing bucketloads of imaginary money. There's no teacher like losing money. I'd say if your penny stock strategy turns over daily it's going to encounter severe headwinds. HFT have all sorts of tricks they employ to cheat you of fills (e.g. "sub-pennying") and force you to cross wide spreads. If you're going to trade that universe, try to minimize the amount of transactions. Aim for higher-confidence trades (so you don't trade more positions than you need to) and longer holding periods (so you don't trade more often than you need to). Also, you'll have to consider the pros and cons of odd vs even lots on market impact and fill quality. These things are all less of an issue if you trade stocks with enough liquidity such that you can "hide in the crowd."

^ what he said!!

@Grant,

Have another go at the contest I reckon. It's hard (to me at least - I've spent the last +3 months at the bottom of the contest ranking) but if it was easy it wouldn't be any fun, right? Worst case you lose/waste some time (opportunity cost), but if it's a hobby that you enjoy it's not really a waste of time as long as you enjoy it and keep learning new things in the process. That's how I see it anyway. You could still trade the $1K on the side as well. :) @ Joakim - Thanks for the encouragement to dink around with the contest again. It would be nice to know specifically what Q needs for their fund. My assumption is that there are ~30 algos in the fund (from < 30 users). So, they must have some sense for what might be a promising addition. Also, I'd kinda expect that Q would be doing some research of its own and then passing it along to the community as guidance. I've been out of touch for awhile, but I'm gathering there has been nothing published. I suspect there is a dichotomy between unlicensed and licensed users, and more specific, actionable information is flowing to the latter (this would make sense, since more of a relationship has been established, along with contracts, NDAs, etc.). Skimming over the forum, there seems to be very little collective enthusiasm in the public domain. Working in a vacuum, staring at a screen in isolation is not so motivating. And since I can't post to threads other than my own, I don't exactly feel welcome and engaged. As I mentioned above, I've been communicating privately with Dan Whitnable on this kind of stuff (and one of my recommendations to him is to have a public forum on such topics). We'll see where it goes. Grant, not sure if this will be helpful but I use a strategy based on this https://h1holdings.files.wordpress.com/2013/03/darwin-adaptive-asset-allocation.pdf @Grant, trying to identify short squeezes might be fun with your$1k, if you’re willing to gamble a bit with it. Short squeezes can sometimes shoot up by X00%. See http://shortsqueeze.com/. The danger of course is that they don’t squeeze, and you’re stuck long on a great short play. Might be difficult also to know when to exit after the squeeze too. I.e is it done squeezing or has it more to go?

Not a recommendation obviously, but might be fun to investigate.

@Joakim

Do you have have any info on how one could use short interest data to write trading algos? From what I understand, rank based on highest short interest and then pick the top few percent. Thanks for any information!

@Arun,

That might work as a short factor, after ‘flipping’ the factor with a minus sign, as you probably want to Short the highest short interest, and long the lowest.

A short-squeeze factor I’m not sure how to do. Maybe together with recent (minute) price movement to get the last end of the squeeze?

Thanks @Joakim

I wish we had access to historical short interest and borrow rate data in Quantopian.

I've been manually trading short squeezes. Over-valued stocks with high short interest and high "days to cover" can tend to squeeze if they pop on an earnings surprise. They may spend the next month steadily drifting upwards in defiance of gravity. The trade I feel most comfortable entering is shorting the stock once the squeeze loses steam.

Thanks all for the feedback! Glad to see that I stimulated some discussion of mutual interest (and for Q hosting the discussion, even though it is orthogonal to their core business, if not antipodal).

One approach would be to get up and going with some sort of ETF-based algo. However, if I do a screen on https://finance.yahoo.com/etfs/, there are no ETFs under $5 per share, and only two under$10 per share. I wonder why there aren't more low-priced ETFs? And why the market doesn't support fractional shares (although I see that there are a few "brokers" who do...I've inquired with Alpaca, just to confirm that they only support whole shares)?

I guess one low-risk approach to get up and going (to learn the ropes, and to establish the technical infrastructure) would be to build a portfolio that attempts to track SPY, but with penny stocks. Then one could morph it into something a bit more tantalizing* by tilting it in various directions away from the market. I'd expect that Q might be useful in sorting out how to do this (tantalizing: tormenting or teasing with the sight or promise of something unobtainable).

By the way, I came across https://www.webull.com/. It looks to be similar to Robinhood, with an emphasis on a smart-phone application. This kind of offering, in my opinion, is ridiculous. I have to think that these companies know exactly what they are doing by offering "Game Boy" legal gambling, and behind the scenes there must be some way they profit to the disadvantage of the game players (but maybe I'm just getting cynical in my old age).

".... probably just luck" . Yeah sure., Grant! So you ARE pulling everyone's chain. But its fun. Thanks :-))

Anyway, now you have a princely sum of $1k to do with whatever you want . The possibilities are endless ... well almost. Your stated constraint is "trade it", and of course you understand that at this level of "trading capital" it is likely to be a gamble. Taking an Engineering approach, have you considered the full available spectrum of Risk-Return profiles for this capital? Assuming that you want to maximize your Expectation Value = Size of Payout * Prob of Success (POS), what is your personal preference? eg. A) "High Risk (low POS) Speculative Venture" with big payout if you win, for example single$1k ticket in a lottery or perhaps something even riskier ......?
B) "Low Risk (High POS) Conservative Investment", for example a capital guaranteed bond fund that accept $1k minimum subscription? C) Something somewhere in between A) & B)? D) "Balanced Approach", with a mixture of 1/3* A + 1/3*B + 1/3*C ? E) "Small Business Venture" approach, e.g. set up an ETF-based algo, just as you mention? F) "Entrepreneurial Venture" approach, as per E) above but offer to sell shares in it to other Quantopians? G) "In God We Trust" approach: Donate it all to charity and see if it indeed comes back to you "many-fold" as promised? Is that trading or a gamble? H) "Find a Savvy Political Advisor" approach: Write to Mr.Trump, pull HIS chain a bit just for fun, and ask him what HE would recommend as a trading proposition? After all presumably even he had to start somewhere, and look where he got to! ;-)) Hi Grant, With a limited$1K in investable funds, I would throw it all in to my wife ( or partner). Buy her a nice dress or a special gift then take her out to an expensive dinner. The downside risks are almost nil but the upside potential maybe enormous which can lead to a happy smile for quite sometime! The Q Sharpe ratio could be infinity unless of course she didn't like the dress or gift and dinner was lousy! I guess you can't totally eliminate risks but can at least mitigate it.

:) @James -- lets trade wives.

Some 3x ETFs have a lower per-share price than SPY, and they are super liquid: https://etfdb.com/themes/leveraged-3x-etfs/ When I was reading up on the Kelly Criterion, my take-home was that on the stock market essentially you should be using some leverage to maximize your returns.

For some reason CEFs tend to have a much lower per-share price than ETFs and ETNs. SPXX QQQX DIAX BXMX etc. Beware of the wide spreads and lack of liquidity on many of these though. I once had an idea of trading based on the fluctuations in the premium/discount on CEFs, but I think my conclusion was that you'd get killed on the spreads. I think there's a lecture by Earnest Chan on the Quantopian YouTube channel where he talks about a CEF trading strategy, but I can't remember which one it is. If I remember correctly the crux of the talk was that the granularity of the backtest data might make such a strategy look profitable when in reality it is not so much so.

What are you thinking in regards to ETFs -- market timing? or some sort of sector rotation? or what? Seems to me going the macro route would be a more difficult task than finding inefficiencies in individual stocks. I want to clarify my previous posts -- I don't think you were wrong when you suggested going for stocks that are ignored by analysts and hedge funds. Rather, I think you're right -- I'd expect the best opportunities to lie in a liquid small- to mid-cap universe. I was just suggesting that from my experience a liquidity filter is very very very important.

@ Tony - The rule I have for myself in this endeavor is to start from scratch--$0. Since I only made$1K in the contest, that's all I have to work with. At this point, I'm thinking along the lines of a "null" algo that just attempts to track SPY, but with stocks priced <$5. I have no idea if this is feasible. @ Viridian Hawk - no thinking behind ETFs, other than that there could be some diversification, and maybe one could start with a very vanilla asset allocation approach. Response from Alpaca: Alpaca Support (Alpaca Securities LLC) May 1, 10:08 AM PDT Hello Grant, We currently only support only whole shares. Sincerely, The Alpaca Team [email protected] https://alpaca.markets @Grant A few algos I like for smaller sums of money: 1. Ray Dalio's All weather portfolio 2. Gary Antonacci Dual momentum 3 Joel GreenBlatt Magic Number investing I've not traded my money on it. Only paper trading. As for fractional shares, motif, loyal3 and folio used to allow it. Thanks Arun. It would be great to be able to roll the$1K into the Q fund, to be more of an active participant in this experiment. Presumably, this would get me information on its performance, along with the private placement memorandum (PPM). I'd gone down this path before with Q (see https://www.quantopian.com/posts/1337-street-fund-aka-q-fund-prospectus), and others have inquired, but to my knowledge, the Q fund is closed off to the user base (I'd think there might be some way around the "qualified investor" regulatory impediment--shrewd lawyers with sharp pencils can do just about anything, as far as I can tell).

I think my next step is to delve deeper into Alpaca.markets. As I recall, they don't host algos (like Q used to do). I would need to set up my own infrastructure (hardware and software)--should be an opportunity to learn a few things.

Hi Grant, this may be the nexus: Quantopian+Zipline+Alpaca to market to contemplate forward.

Think very slowly with $1K to be extreme with capital preservation.. in razor focus with the insidious effect of drawdown: If a position lost 20%, the remainder 80% must recover +25% to return to pre-drawdown footing. If a position lost 33%, the remainder 67% must recover +50% to pre-drawdown. If the position lost 50%, it's virtually game over - Viridian Hawk's proverbial lost shirt. Lost. Bare naked. One could agree with the oracle: Rule No.1: Never lose money. Rule No.2: Never forget rule No.1. Note that Jamie just posted an example of how to filter on ETFs, using some new-fangled functionality. @ Karl - Another good Warren Buffet quote: "Risk comes from not knowing what you’re doing." One advantage in my participation in the Quantopian experiment is that I'm pretty confident that I mostly don't know what I'm doing (if I did, I would have a lot more than$1K to show for it!). I'd say it's not really "risk" if one knows that a draw down of 50% is possible along with some sense of its probability of occurrence, and it is understood to be the nature of the beast (in the "model"). As far as "game over" for a 50% drop--not really, since one can just hit the reset button and re-start with $500 (i.e. just put on another shirt). Of course,$1K X (0.5)^N, where N is a large number eventually ends in game over.

Some related questions:

1. Why do stock markets not support fractional shares?
2. Why don't companies try to keep their share prices relatively low (e.g. with splits, for example, companies could try to maintain a share price of ~1, which would make ownership/trading realizable by a much greater slice of the population)? Technically, these things would be easy (I think), so there must be some interests keeping things at the status quo. Or maybe per share prices kinda naturally align with the25K limit required to avoid pattern day trading?

Regarding your first question, some brokers do allow fractional shares. I used to have a TD AmeriTrade account with DRIP (Dividend Re-Investment Program) that gave me partial shares when the dividend was not enough for a full share.

Regarding the second question, most companies do split their shares when price per share gets 'too high' to allow for more trading, increase liquidity, and perhaps allow for better price discovery. One famous exception though is BRK.A, Buffett's company, which hasn't split the last 50 years. I believe his reason is because he wants to encourage long term investment/ownership of Berkshire, rather than short term trading/speculation.

1. I think the market is just slow to evolve. We only recently started transacting in cents: https://www.investopedia.com/ask/answers/why-nyse-switch-fractions-to-decimals/
2. I think there are different reasons. For companies like $AMZN and$BKNG they've been burned in the past by doing a split right before a market crash, sending their stocks into the low/single digits, which isn't a good look. Generally share prices under $5 are thought of as a warning signal that the company may be in financial trouble. It's naive, but a generally accepted guideline. Spreads also can't get as tight on low-priced stocks that are priced in cents, which is why QTradableStocksUS discludes stocks under$5. Another reason I've heard is that making stocks more accessible to retail may add volatility to the stock, as retail investors ("dumb money") are presumably more fickle and prone to bad trading habits such as chasing and panic selling, driving the share price irrationally. Share price volatility is a distraction for companies that offer stock options as reimbursement to their employees. Also, I think since there's no shortage of demand for $1000/share there's not much motivation to do anything to increase accessibility for the stock. There is not a whole lot of evidence to suggest that stock splits matter. Finance professors have examined stock splits and see no actual impact on a company’s value or performance. Many companies prefer to avoid splitting because they believe a high stock price gives the company a level of prestige. A company trading at$1,000 per share, for example, will be perceived as more valuable
even though its market capitalization may be the same as a company
whose shares trade at $50. Smaller companies may also wish to avoid stock splits because of a danger of share values falling too low. There have been cases where companies have split shares only to see the stock market dive, pushing shares below$10. Psychologically, this may turn off some
shareholders, and in extreme cases, share prices may be too low for a
company to be listed on an exchange. Companies will avoid splitting to
protect themselves from this possibility.

If the position lost 50%, it's virtually game over - Viridian Hawk's proverbial lost shirt. Lost. Bare naked.

Typically returns are commensurate with risk, however a lot of what we do here is about improving our risk-adjusted returns. Therefore, an algorithm that loses 50% is likely taking enough risk to make that back in stride, and then some. So while losing 50% on a faulty overfit algo, to unaccounted for frictional costs, or on fees may mean game over, if you lose 50% because the algo takes smart but risky bets, it's might not be so difficult to recover.

"Risk" reminds me of a book I read awhile back:

Against the Gods: The Remarkable Story of Risk
Peter L. Bernstein

Kinda interesting, as I recall. Up until recently in human history, we really didn't know how to think about and compute probabilities. Humans have really bad brains for uncertainty, as far as I can tell (lizard/monkey/human...it is a mess). Ray Dalio has some good material on this topic, too...emotion versus intellect, with a struggle between the two. Meditation is supposed to help...haven't tried that yet.

I’m thinking of a trial case which should be pretty tame. A homegrown index fund, ~100 stocks, < $5 per share (average ~$3 per share), with beta-to-some index ~1. Once all the kinks are worked out, then add some “smart beta” to see if the index can be beat. Rebalance weekly perhaps. We’ll see if I have any time at all; pretty busy along other dimensions.

Apparently adding some basic fundamental/value filters is a fairly consistent low-level source of alpha in this universe:
https://www.quantopian.com/posts/quantitative-micro-cap-portfolio

I suggest short Uber once it comes online. I made money with shorting Lyft. So easy guessing that cab companies dressed as technology companies aren't worth 100B$. Sorry guys if this is not really inline with algo trading or very generic in nature. It is just that the timing is right. /L *> I suggest short Uber once it comes online. I made money with shorting Lyft. So easy guessing that cab companies dressed as technology companies aren't worth 100B$. Sorry guys if this is not really inline
with algo trading or very generic in nature. It is just that the
timing is right.*

Cant short it for the first month after IPO, the underwriters of the IPO are not allowed to lend out shares for short sale for 30 days

Well, looking at my previous post and the current trading of Uber and Lyft shows that I was correct. Also, I was able to short Uber on Monday on IBKR and Lyft on the following day of IPO. Maybe for large institutional shorting.