New Stock Strategy: US Long-Short Automated

I just started a new US Stock trading automated long-short trading System:

Please find attached 10-year backtest (2009-2018) Equity curve and Monthly results:

This is the result of backtesting 1000 US stocks for 10-year periods and URBN came up as one of the best. One investment unit will be 800 shares of URBN.

Min Capital is $50k recommended or $30k for aggressive investors.

Currently trades will be transmitted via my Tradestation through platform transmit. I only have $19k in my Tradestation Equity account so I will not be able to take all trades due to Pattern Day Trader regulations but I will do my best to follow as many as I can. Therefore platform transmit will work through my sim account.

Subscription will be free temporarily (first subs that autotrade it will get at least 1 free month coupon).

This is not the holy grail, longer term investors are preferred.

Could you please clarify the following:

  1. Is this a backtest of all 1000 stocks in a single portfolio or URBN only?
  2. How the 1000 US stocks were selected?
  3. Is this a walk-forward test or optimization results on 2009-2018 period?
  1. This is the backtest of URBN only.
  2. The 1000 stocks were selected on the basis of liquidity and a Stock price of between $15-$95 on NYSE and NASDAQ stocks. To save time I first conducted a 1-2 year backtest (filter) and if the performance was favourable then I went on to conduct the full 10-year backtest using 750,000 minute bars. The whole filtering, selection and backtesting process took around 6-8 months to complete.
  3. This is not a walk-forward test, it is a backtest using Tradestation Strategy Backtesting Tool.

Please let me know if you have any other questions.

  1. Did you select 1000 stocks only once at the end of the 2009-2018 period?

  2. Sorry, I am not familiar with the Tradestation Strategy Backtesting Tool. You probably have some parameters in the system. Did you find the best parameters over the whole 2009-2018 period and then use them to run the backtest over the whole 2009-2018?

  1. Yes I only selected the 1000 stocks at the end of the 2009-2018 period.

  2. All the parameters (or inputs in Tradestation EasyLanguage) are and have been the same, there has not been any changes made for any of the backtests. So no curbfitting took place.

I hope this explains it but happy to answer any more questions.

no more questions. Thank you!

1 Like

Just exceeded $1 million AUM.
Fortunately this is 100% automated so it won’t affect the strategy whether it is $50k or $100million is being managed.

Your strategy didn’t make any profit between Trade 1250 and Trade 1750 (roughly), neither did it between Trade Nr. 1900 and Trade Nr. 2300 and between Trade Nr. 2500 and Trade Nr. 2900. Do you really think that anyone would psychologically be strong enough to trade a strategy in which you need roughly 500 trades to make a new high? What are the timeframes of those periods without new highs? How much transaction fees did you take into account in your backtest? I assume 0.

You are wrong in all points. One actual full trade consists of 3 trade numbers in this backtest, 1 is trade entry, second trade number is first target and third trade number is second target so you will need to devide those numbers by 3. These 2 sideways periods lasted around 6-8 months so out of a period of 120 months a 6-month period is 5% sideways equity movement and 95% continuous equity run-up period. I don’t think a 90 or 95% period of continuous profits is such a bad performance.
I had two slightly losing years (almost break-even) since 2009. 2008 was actually a winning year so when the world economy collapsed this system made money. A long short system’s robustness is shown by how it performs during market crashes.
Commission is counted at 1 dollar roundtrip per 100 shares.

In this screenshot it can be seen how it performed around the time Lehman collapsed (starting Lehman crash preceding September 12th 2008 till September 30th).

This is not the holy grail, it is for long term investors. There are many systems for those seeking quick bucks, martingale and the like so I am not competing wth those gamblers, this is 100% automated, no human emotions involved at all.

Several people asked for backtesting results for 2008 when all markets posted a loss.
URBN for 2008: