Post Mortem: What Went Wrong

Trader's Mindset

May 01

Recently, I took some new methods I had been trading in a realistic simulator (Rithmic) and went live. My simulated results are shown below:


While,  I did have some larger losing days then I would have liked in the simulator: overall I thought the results were promising. As an aside, I didn’t actually lose more then ~($1,000) in any single session. The sessions where it shows I lost more were because I had already closed profits in excess and because Tradervue looks at the session differently.

I like to tape read markets and predict or anticipate them. For these results, I traded with tight stop losses and placed many trades per day primarily in the S&P 500 E-mini ES. I used my tape reading software and some custom graybox assistance tools to help me.

While, I had some experience trading discretionary and systems, this would be the first time I would attempt to trade with a tight risk per trade and in a more frequent style. Unfortunately, I made several mistakes in my live account and suffered extensive losses right off the start. The losses were significant enough such that I am returning to the simulator until I have corrected the problems and have stronger evidence I can make this work.

So, what went wrong? In order to understand what went wrong. I must explain what I was trying to do with this method. Most futures systems trade selectively. They trade infrequently and they wait for conditions to align, typically abnormal conditions, and tend to make large bets. This contrast with how high frequency trading works where those systems tend to make small bets and trade more frequently and avoid the abnormal markets. My goal was to develop a method where I could express my market perspective with a strict and limited risk per trade, find enough trades that I could be profitable most days, and do so in such a way that  I would have high probability of making those returns. This contrast to taking a larger tail risk, i.e. a large risk per day and waiting for trades to recover.

Actually, I feel like I was close to making it work. The ability to trade this way requires ultra precise entries: which I am pretty good at. But, that said, I knew my stop loss settings were somewhat arbitrary. In the simulator, I wasn’t risking much. I was risking something because it was a paid tryout. And, I had enough confidence that I just did not care that my stops were poorly chosen. Even with precise entries, you have to know whether to hold the trade or not. This is very difficult when trading with tight stops because it is not enough to pick the direction but whether or not you can hold a trade depends on the very specific market conditions.

Going live, I was able to get similar entries, that wasn’t my problem but my knowledge that my stops were not set optimally, and it was something I knew I needed to work on led to a series of mistakes on my part. The first few days I thought we had entered a low volatility environment, and I knew that my techniques might not work in that environment. So, I adjusted my stops but I didn’t adjust my contracts down correspondingly. I quickly had larger then anticipated open losses. The other risk with tight stops is having a series of correlated losses. So, this was even a problem in the simulator and in order to avoid that you have to keep your win ratio high and avoid taking too many trades in short sequence.

I do think that discretionary trading can outperform most systems: the cost for that performance is operating more on the hard right side where it is less possible to ascertain if you have an edge or what the edge is worth. The more discretionary, the more frequently, and the more leveraged that one trades then the more difficult it will be to translate good simulated results into live results.  Every single factor makes it more difficult.  The combination I chose was the most difficult and it turned out to be a bit too difficult to trade in practice.

The initial changes and corrections I will be looking to make in the simulator will be going with an automated quantitative based stop loss and first target and then managing the second target with my discretion. Normally, I can hit good numbers on my first few trades. So, I think if I can develop some stop losses that I have confidence in and just post my normal numbers on the first trades that I may still be able to make this work. But, before risking and losing any more money in the live markets, I will be working out the problem areas on the simulator and seeking verification I can make things work. I strongly dislike going back to the simulator from live trading because obviously I miss out on any recovery. However, the reality is for myself, at least, futures trading is highly leveraged, and I cannot afford to learn in the live markets with futures. The costs are just too high: it is far better to work out problems without any risk on the line.

About the Author

The author is passionate about markets. He has developed top ranked futures strategies. His core focus is (1) applying machine learning and developing systematic strategies, and (2) solving the toughest problems of discretionary trading by applying quantitative tools, machine learning, and performance discipline. You can contact the author at