Some 10 years ago i`d`ve gotten the piece of your $10MM in no time.
System that works but lack the capital, please advise
- Thread starter longshort
- Start date
Some 10 years ago i`d`ve gotten the piece of your $10MM in no time.
CFD has too much slippage.
As others have asked - if you are actually confident about the system, ask money from friends or family. But since you won't, it makes me think you're not really confident about the performance. $25k in Netherlands isn't a lot of money at all. I understand if you were living in Northern Siberia in a small village then yes, $25k might be out of reach.
Obviously he's day trading stocks, well swing but needs the option to close same day occasionally, hence 25k+.
Stocks trend better than indexs imho, i did very well till 25k rule came in 15years back, still pissed about this rule, been unable to raise 25k+ so can relate to his issue.
Need made it with options but blew up badly, forex from $300 huge waste of time for like 13years.
Ohh how the years fly by
My first thought as well. His backtesting begins at the very point the market turned.
How convenient.
Ha ha. Lol. Netherlands indeed. Ok, longshort, come by and run this baby through me and see what we can do about it. However:
1) Is the system a purely intraday play? It surely looks that way.
2) If so, the pnl curve looks way too smooth for only 20ish intraday stock trades. "Look-ahead bias" jumps straight at me in this one. Check your entry/exit price assumptions to start with.
3) 14 cents per trade on average for an intraday system in the US stock market these days with the VIX permanently in the low 10s looks like a pipe dream to me, however. You must have found a very dark and overlooked corner in this market!
Enjoy!
No delays modeled and no tick data used. It's low frequency (not scalping) on multi-minute bars. Yes, the same $50K lot size every time, although both sub-systems can have an overlapping trade, which would somewhat correspond to strength of the signal.
First, I assumed 09:30 to 16:00 New York time represents 100% of the market, since this is the only time when positions are permitted. Using that standard, the more active sub-system is 33.7% in the market. The less active sub-system is 19.8% in the market.
You then bring up a good point. I agree, it's useful to test how much performance is due to market going your way. So I simulated a buy-and-hold system that is always long but only in the permitted time from above. Using no slippage/commission, in several years of the data the buy-and-hold system is significantly positive. In other years, also several, buy-and-hold is significantly negative.
So on the one hand, there are alternating bull and bear markets in the underlying data. On the other hand, both of my sub-systems only have positive returns in all years of the data, after slippage/commission.
Interesting method, never heard of it done this way. I like to look at edge strength and number of trades. Profit factor >2.5 and 300-500 trades can be significant. Profit factor 1.4 and I'd like to see 2000+ trades. The system complexity here is low in my opinion and the parameters don't really do a lot. Easy to read the code and understand the logic.
Please see my above answer, there are bull and bear markets in the backtest period. The backtest uses all data since instrument inception.
Not ideal but looks feasible. Thanks for the idea.
It's easy to substitute $50K initial capital with $200K, which reduces all percentages in the report that are based on initial capital, to one quarter. Instead of 20% drawdown, you now have 5% drawdown. Easy to reduce leverage because report is without compounding. The returns in % decrease correspondingly to one quarter. Year-to-date return 2017 is now 11.83% instead of previously 47.32%.