austin.. what would be a reasonable fwd test period. 2 weeks 2 months 2 years ?? my system is more swing than pure intraday. ~ 2 trades/ day. if it was pure intraday, (not sure how many trades/ day it would require to classified as such), i assume the fwd test period would be less due to the high frequency? thanks for the the posts here, bb
It looks like a simple "buy and hold" over the same period would have returned almost $50,000 per single contract (no compounding). I am not trying to put down your system (it certainly has a better risk profile than a buy-and-hold). Just pointing out how difficult it is to beat the market. It's a lot of sweat to run the ATS for 5 years, pay $13,400 in total slippage, and realize only a third of the market return.
<i>"It looks like a simple "buy and hold" over the same period would have returned almost $50,000 per single contract (no compounding)."</i> Very true. Who would have known when to buy, and when to sell? I've heard the same thing from people outside the trading world many times in the past. Here's the question I pose to them: "When is the next time to buy or sell?" Please take a peek at the GBPUSD charts tonight, and let me know where = when the next five-year buy(sell) and hold signal comes. The system knows what to do in that time... do you?
That's a good guess, although there is no "queue" in Frost's case, because the bot uses market orders. Or, more precisely, the queue is very short. As we all know, it normally takes less than a second to fill a market order with IB, especially if the order size is 1 contract.
No, I don't. I would in fact take the system with this profile and trade it, too, just like you did. I am simply pointing out how difficult it is to develop a consistently profitable mechanical trading system whose P&L would beat the market.
I don't intend to hijack frost's thread at all... that type of stuff happens way too often in this forum. Frost, the variance between your IB walk-forward and real execution must be somewhere in the translation of real money execution = slippage versus demo mode. Based on what you say, there can be no other reason. I wrote many systems which legitimately averaged $50 per trade over five years' time. But... back out -$10 per trade slippage on each turn and -$10 per trade bid/ask in the ER, we've lost -$30 of the +$50 hypothetical profits. That will change the equity curve in dramatic fashion, unbeknownst beforehand. From what you say, IB's difference between sim and reality is the chasm of death for your bot. If your system averaged $120 per trade, it would likely hold up thru the difference. That's one of many reasons why pure intraday bots designed by retailers like us won't hold up... the big pros are driving Ferraris while we sputter along with Hyundais. <b>Steve46</b> has earlier mentioned the difference between retail and institutional bots... light years apart. That's pretty much all I have to share on this topic here. I found system trading to demand far more time, focus, study, work and emotional drainage than discretionary trading. I learned a tremendous amount about how markets really work thru my time spent in bot-world. It might have made the difference between success as a discretionary trader and failure for me, in all honesty. Some of the surprising lessons learned building & breaking down systems in years past are foundations of how I view and approach market action today. No regrets, just a path I've long since left. Best wishes to all those who prefer that style of trading
<i>"I am simply pointing out how difficult it is to develop a consistently profitable mechanical trading system whose P&L would beat the market."</i> capiche`! You can sure say that again... bot building is by far the toughest thing I ever tackled per trading. IMO, we retail traders arm ourselves with proverbial sticks & stones with software available to us. A relative few gifted, highly competent individuals take those primitive weapons and beat market action to submission with them. The rest of us are grossly misarmed. That's pretty much my view of the whole bot thing. Big pros have much more, far more sophisticated tools than we do. Not a level playing field, at all.
Maybe, however I don't really believe "they" are smarter than me. As for tools, you must play the game in the sandbox that makes most sense for where you are coming from. It seems to me that if you don't have lightning fast software, then you must expand your time frame. Then even super-quants should have relatively little edge over you just in terms of software and execution.
<i>"Maybe, however I don't really believe "they" are smarter than me."</i> If the Native American Indians had equal weaponry as white settlers = invaders did, this country would never have been taken from them. Many of the indian leaders were absolute military geniuses, and far more wise & intelligent (in every way) than their European adversaries. There was no comparison of armament, hence the indians' demise. == The longer a timeframe systems are built to trade, the more level a playing field exists. Back off from the intraday = scalper bot world, work in the swing-trade world and retail traders can compete. The only reason I don't trade my FX bot is inefficiency. I can greatly outperform its +180 pip monthly average over lifespan as a discretionary trader in emini markets. But, the fact remains this bot has been profitable every year for the past five consistently. Those who are interested can use TS canned "volatility expan open" strategy to code something very similar to what I did. No mystery there... start with a 60min GBPUSD, EURUSD or GBPJPY chart and run different optimization tables. Make sure the parameters for longs and shorts are balanced on both sides, optimize one hard stop and one profit exit together. That's it... end of recipe for overall swing trade bot success. Signed off