How about if backtested, and better results if skip sometimes?
I have a system where the 6 years backtest shows that if I skip every trade after a winning one, I get better results on the long run.
This means that if I followed the theoretical system trades on paper, I should wait until the loosing trade, and then enter with real money only the next signal, after the loosing trade.
I thought that this is an anomaly by accident, and probably because of "curve fitting" effect,
After 1.5 years of live follow-up, AFTER the 6 years backtest - which means that this 1.5 years is an independent period - I still see the same anomaly.
Could it be that when sometimes, the index behaviour fits well with my system, and sometimes not - this change in behaviour may be visible in the trading results, and by skipping some of the trades, I get better results?
Then I really should add into my system, a feature that "enter the trade only if the previous trade was loosing."
The famous "Turtles" did something similar. I can't remember now whether they skipped the next entry after a winning or after a losing trade, but one of those parameters was part of their original education and instruction. I never understood it at all: it looked like just a variation of the "Gambler's Fallacy" to me, but it was certainly taken seriously, at the time. I suspect it was a backfitting/curve-fitting anomaly, of some kind.
The following user says Thank You to Tymbeline for this post:
The Turtles used two breakout systems, one was 20 days and the other 50 (or 55 - can't remember exactly).
Only the short-term 20 day system skipped trades. The longer-term system did not skip trades if the previous trade was a winner. The logic was that on the short-term system you were much more likely to be chopped around, while the longer term system ensured that you got into decent trends that could be missed by a skipped signal on the shorter term system.
The following 3 users say Thank You to grausch for this post:
Is this thread for real? This reminds me of people I used to meet in Vegas who had the same notions. "If I get ahead this much..." or "If I win (or lose) this many hands in a row....." or "If I only play (X) amount of hours per day...".
Please understand that if you are trading a setup that has a positive expectation--- then you take every single trade that fits that criteria. Any statistic (or crazy notion) that you may come up with that leads you to believe otherwise is not sound logic. These trades don't know how much you are up or down, how many in a row you've won or lost -- or -- whether or not they are even taken during the same trading session.
Now if you have trouble going on tilt and sticking to your own plan, then I have no answer for you other than QUIT. I'm about sick of seeing foolish, unsound advice being tossed out as acceptable -- in order to coddle people who have discipline issues, trade too large, or too frequently, do not grasp basic math or know how to keep simple statistics.
Positive expectation trades are profitable over the long run. Take them! Negative ones are not. It doesn't matter how you click the mouse, slouch in your chair, walk away from the screen, how many you have won (or lost) in a row, whether you have argued with your spouse that morning, your kids are sick, how your confidence is running. None of this matters if you have made a trade with a positive expectation.
The following 3 users say Thank You to Paige for this post:
I agree with you that people are sometimes coddled a bit too much when they have no discipline or blow up due to too much risk being taken.
However, unless you know the basis for your expectancy number, then taking every trade that meets the criteria may not be the best option. For instance, a fx system that takes breakouts around the time of the London open might do well, but if that same breakout occurs when just the Asian market is open, it may be better to pass on the signal. A long-term trend-following system will do well taking all signals when market volatility is low, but in highly volatile times, it would be best to cut back exposure or to skip trades completely. However, I do not agree with skipping trades purely because criteria like the last trade being a winner. There are better ways to determine whether or not trades should be taken.
With regards to positive expectation trades, unless you are the casino, i.e. with a nice fixed and easily (or not so easily sometimes) calculable edge, then some discretion is required about when and how often to play the game. In the casino, you are playing in a closed environment and the casino should apply its edge as often as possible, whereas in the market outside factors may have a large influence on your perceived edge. No matter how well you calculated your expectancy number, you can never account for the unknown unknowns, nor can you ensure that the conditions that led to the positive expectancy persist.
Therefore, taking every single trade that fits the criteria without being aware of the current market conditions is also not sound logic.