Back to Community
The returns are somewhat unexpected

This algo needs to be optimised so as to eliminate the excessive draw down, I don't believe it would survive live trading otherwise. I am curious to see the results on other commodities. What sort of results are others getting with this template?

Clone Algorithm
Backtest from to with initial capital
Total Returns
Max Drawdown
Benchmark Returns
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
# Backtest ID: 55ad36c05214020c67050631
There was a runtime error.
3 responses

The first transaction buys $300,000,000 in securities. Something tells me 300,000x leverage isn't allowed (anywhere I know of at least).

Hi Spencer. firstly, I don't see this transaction, or what you speak of in the data set...the first purchase is for 100k, I don't think that this equates to the amount of leverage you mention...1lot on average is is 100,000units, and the requirement for the CFD component (difference) in pips requires around [email protected]$200.34. I did a lot calculation on the largest holdings, an it worked out to be around 70 lots, when the account was at $868,122,178.28

$200.70 qty 6,980,000 Note: /100,000 = 69.8 lots

Position $1,400,886,000.00 Gains $483,965,280.00
Cash ($868,122,178.28)
- -
total positions :$532,763,821.72 total gains: $483,965,280.00

From past experience I have maintained >1m of positions using only <$500 capital

This would appear to be the effect of compounded loading and interest combined with leverage...though I could be wrong
Anyone here able to validate/clarify..or offer some useful constructive criticism?


Spencer is right -- if you run a full backtest and click Transactions, you will see it purchases 10,000 shares of SPY almost every minute (over $1million per minute). You are purchasing real shares for the full amount (as far as I'm aware, there are no CFDs on Quantopian)

        if data[context.spy].price < avg_price*.95:  
            if momentum:  
                order(context.spy, -10000)  
                log.debug("Selling spy -10000 momentum")  
            elif reverting:  
                order(context.spy, 10000)  
                log.debug("Ordering spy +10000 reverting")  
        elif data[context.spy].price > avg_price*1.05:  
            if momentum:  
                order(context.spy, 10000)  
                log.debug("Ordering spy +10000 momentum")  
            elif reverting:  
                order(context.spy, -10000)  
                log.debug("Selling spy -10000 reverting")  

You could try order_target_percent instead, but be aware that your order takes time to fill (quite possibly >1minute) so you should probably wait for the current order to fill before issuing another.

if not get_open_orders(context.spy):  
       order_target_percent(context.spy, 1.0)