Apex, who's your broker ? Buying limit orders on lows of bars and selling on highs is a difficult job...Neither TS, nor OpeneCry have such fills in my experiment... Seriously, if somebody knows a broker that permits it, I would grant him bottles of good wine from my country( France )... Frosty, I've made some experiment on "when to stop a system" lately. I did it through tests on different after-release systems. some were consistently making money, some were incredible losers. I found that trading the equity curve was not necessarily increasing profits. The best thing is to define a maximum level of loss from your all-time equity high( I use 2 initial MIDDs, but you can use 2 standard deviations...). A thing to avoid is the moving average to define when to activate / desactivate a system( like trading when the equity is above the average... there's too much activation / desactivation ). An other thing that is quite relevant is to define channels around your equity and to begin trading when the equity is below lower band and stop when it's above upper band... Just some random ideas... If it can help you find a systematic way to face your drawdowns... BYE
tradator, Thanks for the advice.... I like the idea of using channels on my PL chart to decide when to activate... For all those wondering the system made +$10 Very uneventful it seems....
The system netted ONE TICK today in eRL ? This, after a mombo, straight-up trend in the morning ? Yes, it was fast, but STILL ! Granted, the afternoon was choppy up till 3:30, but then the floor fell out....but the market was showing it's (overbot) hand by then. Oh, I would definitely pull the plug on this guy till you can determine if it can be profitable at all in this market.
u are down almost 1k bucks and keep in mind that it aint more cuz your bot really pulled off the first day it went live...that was prolly a once in ages thing...u better pull the plug on that babe and make him behave before hell break lose, innit.
First month in the books and I am down a total of -$751.... Concerning the first day, that is DEFINETLY not a rare event... past month it seems like it has been.. but it is generally not rare at all... With that said I am considering but havent made up my mind yet about pulling the plug just until monday. I started working on a "chop detection" filter.... and early results are looking good.. but too early to tell yet.... I might want to wait until monday to see where this filter takes me...... if it keeps looking like this it will blow my current strategies way out of the water....
Frosty, I assume your system is based on time series with constant time bars. Have you considered constant volume bars ? I have no idea if it would make things better/worse/no different, but I think it would be at least educational. Trends certainly look different on constant volume charts. See the attached charts of this morning's K200
I have tested this strategy on: 1. Constant time bars 2. momentum/range bars 3. volume bars 4. tick bars Constant time bars seemed to prevail everytime... at least for this particular strategy.... Those other types of charts look great while looking at them.... but the results did not seem very stellar at all when trying to actually apply them....... With that said, most of my focus was on the momentum/range bar charts... I only briefly ran some tests on the volume bars... It may not have done great because I did not spend the time to find an optimal volume to use... Will have to revisit one day... also, if I am not mistaken Ib's data feed is not good for volume..... Someone more knowledgeable on that particular issue may be able to point out what the problem is....
To my knowledge, IB feed is fine for volume, but not for ticks. Just use cumulative VOLUME event to calculate bar volume. Do not use LAST_SIZE.