"You will have to prove to yourself and others that your suggestions bring in at least 1 tic per day on average extra profit over the original method rules. "
People (Jack) who can't think or write clearly are not useful in my world. I was not only commenting about this thread or Jack's one post. I have seen a few of his ditties involving my name in the last few days and they are all gibberish. You did not irritate me!
I see. Well, in MY world a martian speaking in morse code is useful IF he is speaking the truth and that truth has an edge. He spoke one clear truth that I did understand. Maybe he has more, maybe not. It's a pain in the ass to need translators, I'll give you that...but you gotta do what you gotta do: Whatever it takes to win. Obviously we inhabit different worlds. Not judging which one is better, because mine is hellish a large percentage of the time.
I have yet to see explained the win/loss SIZE issue. The number of wins and losses are irrelevant, what matters is the SIZE of each. Your "break-even minus commissions" statement is based on the faulty assumption that your average win and your average loss will be identical - that isn't the case. So while you're telling everyone to prove a 1 tick per day improvement, how about you first prove break-even minus commissions.
Since each win or loss has a 50% chance of going the other way, by definition, regardless of SIZE, the aggregate does too by commutative properties of the addition. Of course, you must substract slippage and commissions each day. Of course you could have made the proof yourself had you thought about it for a few minutes, hours, days, weeks or years. http://en.wikipedia.org/wiki/Commutativity So, as you can prove to yourself and others, what you think matters, does not matter at all in reality. Now...if you replace the coin tosser for a "more intelligent" algo. Then yes, size of wins and losses as well as the %W and %L come into play.
try a "modified or semi martingale" approach. If you're winning at least 45% of trades, then such an approach will turn a losing system into a winning system. The only drawback is when you get below the 45% threshold. Actually, there are modification that can allow a 40% win rate with equal r:r to be profitable. If you create a favorable r:r with a modified martingale, then your win rate can easily go down to 30% and be profitable.
I believe Martingale does not work in a random normal distribution, and is lethal in a fat-tailed distribution (like the markets). LTCM was using a martingale-type approach of adding to losers. I think more money has been lost playing martingale than any other "method". But knock yourself out. Remember that: "You will have to prove to yourself and others that your suggestions bring in at least 1 tic per day on average extra profit over the original method rules. "
I think it would be fairly easy to create a solution. The trade as proposed is random. Solution: Introduce a pattern to the trade to create an expectancy. Determine the trend in some way, say by counting up and down days for the previous ten days. Only take trades in the direction of the trend. I could not be bothered to see if this works, but it seems logical that in the long term it would.