Hi Quantopians, especially Delaney,
[I come in peace and i hope you are friendly]
Although I’m still “new around here”, if you will allow me to be so bold, I would like to start my very first thread here on the Quantopian Forum.
It is in the form of a challenge to anyone who wants to play, and I hope it will be fun and educational for everyone. You can do your own scoring for this one. There is no prize, at least certainly not from me anyway. In this particular challenge you can even cheat if you want to, although you will only be cheating yourself, because the aim here is to help everyone win something very special in terms of your own knowledge.
I created this for two reasons. Firstly because Delaney at Quantopian has been strongly urging us to look at “alternative data” and use inputs other than just price. Secondly, I know there are some people who don’t believe in “fundamentals”. They put forward various reasons about why they think “fundamentals are BS”, but I think they are wrong, and I would like to demonstrate that to you now.
Please take a look at the results below, from an algo that I wrote. Those results look rather un-spectacular, don’t they? Yes, I think they are rather un-spectacular too ….. except for one small thing……. The algo used NO PRICE DATA AT ALL!!
The only inputs are from the Morningstar Fundamentals data freely available to all of us, and excluding any ratios that involve the stock prices in any way, for example PE ratio, Price to Book ratio, Earnings yield, etc.
To anyone who thinks that Buffett & Munger’s consistent performance over decades is just a “statistical anomaly” (i.e. lucky ...yeah, sure, just like the idea that, given enough monkeys with typewriters, one of the monkeys will surely write the entire works of Shakespeare), I would say OK, continue to believe whatever you want but, IMHO, fundamentals really DO work and I believe Buffett & Munger are excellent proof of that. So is the algo output shown here….. Unless of course the only reason that it works is because the entire period from September 2009 to September 2017 is just a big bull market, and everyone knows that absolutely ANY fool whatsoever can make tons of money very easily in a bull market, right? ;-))
So here is the challenge for you:
Design an algo to beat the results shown, over the 8 year period from 1st September 2009 to 1st September 2017 (as an equal basis of comparison for everyone), using ONLY Morningstar Fundamentals data, EXCLUDING the price-related ratios. All other constraints are exactly the same as per the real Quantopian Open Contest, including "competition transaction costs" etc, and especially leverage <= 1.
You can score yourself however you want to really, but my personal “scoring system” for this little exercise is as follows. (Please note: this is not intended to have any particular relationship to the way in which Quantopian might calculate the Quantopian Open Contest scores and, as far as I know, it doesn’t but it is still useful, at least to me).
Uncle Tony’s Score = 100*Sharpe* (Returns% / 8years) * (1 + 10*(alpha-abs (beta))) / (1 + Drawdown%)
On that basis, the example shown would have
Uncle Tony’s Score = 100*0.93* (24.35 / 8) * (1+10*(0.03 – abs (-0.04))) / (1 + abs (5.70)) = 54.2
Can you improve on that? If so, please share with us how did you do it?
Remember, NO price data or ratios that involve stock price data in any way. Do your own scoring. This is designed as a learning experience. Have fun!!!
After playing a few times, I hope you will be asking yourself why you or anyone else would ever even consider throwing away a whole lot of perfectly good alpha by NOT using available fundamentals data.
Delaney, just imagine what we could do if we actually added PRICE data as well!! ;-))
Cheers, best wishes, Tony M.
|Returns||1 Month||3 Month||6 Month||12 Month|
|Alpha||1 Month||3 Month||6 Month||12 Month|
|Beta||1 Month||3 Month||6 Month||12 Month|
|Sharpe||1 Month||3 Month||6 Month||12 Month|
|Sortino||1 Month||3 Month||6 Month||12 Month|
|Volatility||1 Month||3 Month||6 Month||12 Month|
|Max Drawdown||1 Month||3 Month||6 Month||12 Month|