Without a live account track record, how do we know if you did not over-fit your data during the backtest?
Well we did not overfit the data... Considering the algo is forward looking... You can even see open positions as of now and how much the open positions have contributed to the bottom line as at yesterday's closing prices... Remember we are not doing intraday... We hold positions for average 7 weeks... So from point of entry you can know since trades are implemented on following day market open once you get the signal... And the price of market open is known.. You can even choose in real trading to implement positions on a lower market price once you know the market open price... Moreover 80% of positions are also closed on market open once you get a signal... And also we have used a round-trip commission of $40 for every contract to factor slippage and spreads... The issue is for me is not if it works... That have already proved and continue testing even in live markets with some trades... The issue is scaling since how the code is developed with limited capital you can only trade few markets. E.g. Based on a notional capital of $100,000 you can only trade 3 eurodollar contracts and some micro contracts in gold or currencies (illiquid).. Even though in reality you have as much as 95% cash... The key thing is risk is small to make sure you have as much cash as possible to hedge...
3 GE contracts on $100K? A savings account has more volatility. It gets better... account held in Kenya! Enough to hedge? Hedge what? You’re going to spread GEs after you’re down? You wasting everyone’s time, but it’s a fun read, so thanks.
Well I didn't say you must put the idle money in treasury bills or here in Kenya.. That is how we would do it... Considering Kenya 1 yr tbill can give me 8% free money...but note that 70cts invested in tbills only account for less than 15% of total return, 85% of return come from trading futures.. E.g return for 2019 YTD (-3.44%), 2018 (25%), 2017 (13%),2016 (17%)...
You can claim whatever you want but in order to find a real investor you will have to back up your claims with real money and real verified results, not with some nice promises and bar graphics every 10 year old could make.
I have news for you my friend. Even 14 years is not sample big enough to test strategy with holding period of 7 weeks. One could easily overfit it, I have done it many times. If your holding period is few hours, than 10 years worth of data might be sample big enough to make it very hard to over-it, assuming it take enough trades.