How do you determine varying the size? Allow me to take a gander... you are saying that if the trade has a higher probability of working out, due to past (positive) performance results, then size is more?
Not quite sure if that's what I mean. If you are to pick the next trade you do and size up 10x, then trading the system with a 66% chance of winning 1:1 is better than only having a 33% chance of winning 3:1.
I think amigasearch hit it on the nail. I believe it's Marty Schwartz who echoed a similiar statement in his book. If the market gives him 4 ticks, he takes it. He needs to ring the register. Ringing the register and taking a profit tells him he's right, and he needs to be right...over and over again. So, it really depends on your psychological makeup. What works for you?
This is true for me, too. That is why I am trading systems with between 65% and 70% profitable, but only about 1-1 ave gain/loss. m
say you start with 10k and trade 10 year notes (margin req. at $800/contract) with the risk at $500 12 trades 33% system ($1500 gain/$500 loss/contract) = max of 2 contracts 8 losers in a row (-$8000) then 4 winners after (+12000) = +$4000 after 12 trades 66% system ($500 gain /$500 loss/contract) max contracts = 3 contracts 4 losers in a row = -$6000 then 8 winners = +$12000 =+$6000 after 12 trades it would be very important to know if these system streak losers and winners or are evenly distributed.
In the system I am trading there are occasional streaks of losers, but no more than 6 in a row in backtests (so probably more in real trading). During the DD periods there are more trades. One way to improve the equity curve is to limit the max number of entries at one time. Smooths out the curve a great deal. Of course, I have to do this for the sake of money management anyway. m
A 20 cents question: Before placing an order for each trade, I would think there should be an inherited uncertainty for the gain/loss ratio (whether 3:1 or 1:1 or else) of the trade since we would not even know in advance whether the market will go south or north for the trade (if every trade would have the same gain/loss ratio as the system average based on historical data). Then the issue would be "how are we newbies able to derive an estimated gain/loss ratio for a particular trade?", assuming we could know for sure the market would go to our favourable (1) Direction (2) First, and (3) Reaching the profit target (4) Before bouncing back and (5) Not touching the stop loss.
I'm not sure I understand your question. My expectation on win/loss ratio is based on 6 years of backtesting and 1.5 years of real time trading. The system has maintained it's 65% profitable trades statistic throughout (though on very short periods, of course, it varies). Without the testing and real time trading, I would have no idea what to expect. m
Therefore, imo, the ratio (alone), which is merely a statistical data with possibly very minimum predicting value, would Not be considered a (critical) issue for decisions!?