Slippage for trend following system

Quote from Zr1Trader:

Here is a backtest of my system with 1.5 tcks of slippage calculated.


From jan1 2006 to present.


Total profit $166,000
-
$39,000 commissions

=$127,770 Net with 1 contract Max drawdown $7,670


Now, what happens if I want to calculate slippage with 3-5 contracts? How do I test for partial fills ect? Anyone use ninja that could point me in the right direction?

Will forward test on sim for 6 months to a year before taking it live.
More...

What timeframes? In my experience I would test the last 10+ years if possible. What you have will break down, usually very quick. Good luck either way though.

I don't know your methods etc. but if using Hour Bars what you have isn't significant, IMO.
 
Quote from Algo_Design_Kid:

What timeframes? In my experience I would test the last 10+ years if possible. What you have will break down, usually very quick. Good luck either way though.

I don't know your methods etc. but if using Hour Bars what you have isn't significant, IMO.
More...

Those last results were on the 5 min timeframe. So according to you it will breakdown? Why would it? Not being defensive just curious and want to learn from others experiences.



Testing it on the daily bars from 2006 to present

$280,172 total profit
$6598 Max Drawdown
184 trades x $4.50 commission = $828

Net $279,344 1 contract

thing is I didnt want to hold overnight and deal with rollovers ect. Because I already have longterm investments. I was looking for something intraday.

I don't have ten years worth of data, only from 2006 but if a long only system can last through the 08 period and 09,10 wouldn't you think I have something here? Ive tried this on different baskets of futures and the results are similar which they should be according to the underlying idea behind the system, Cut losses short, let winners run.

Slippage on the lower timeframes is what my main concern is . My advantage against the big funds with millions in R&D and all that other hockey pockey is that I can get in and out in an instant, they cant. I figured there would have to be some room for profit left for the little guy that can come up with a better system than other small traders.
 
Quote from Zr1Trader:

Those last results were on the 5 min timeframe. So according to you it will breakdown?



Testing it on the daily bars from 2006 to present

$280,172 total profit
$6598 Max Drawdown
184 trades x $4.50 commission = $828

Net $279,344 1 contract

thing is I didnt want to hold overnight and deal with rollovers ect. Because I already have longterm investments. I was looking for something intraday.

I don't have ten years worth of data, only from 2006 but if a long only system can last through the 08 period and 09,10 wouldn't you think I have something here?
More...

I have many things that have just fallen apart. Actually about 99% of them. I work many hours a day, but only have maybe 1% that have kept performing out of all my work.

If you are confident I hope you do well.
 
Quote from dave4532:

Since 2006 on 5-minute bars and only 184 trades sounds like over-fitting to me.
More...

That was for daily bars, the results before that were the 5 min bars.
 
Quote from Zr1Trader:

Bump

Can anybody explain why or why not a system that uses reverse orders is more beneficial than using stop orders?
More...

With reverse orders you are always in the market an you can be hit by a black swan. For me a good system is one that makes the most profit by staying in the market as little as possible, or better say for the least time possible.
 
Quote from intradaybill:

With reverse orders you are always in the market an you can be hit by a black swan. For me a good system is one that makes the most profit by staying in the market as little as possible, or better say for the least time possible.
More...

I'm talking in regards to slippage.... reverse orders vs just plain old stop orders, any advantage or disadvantage? Dissaster stops would be present. But thank you for your input.

I'm wondering if I'm better off with a long only approach or a long short approach in regard to slippage. If I am long the exit signal would be the same thing as a sell signal if i was using the long short approach.
 
Ok well so far I have tweaked a few things to reduce slippage as much as I can . The best way I have found to accomplish this is enter long only and enter with limit orders. By entering with limits and sacrificing missing some long entries but thats ok. The benefit of knowing this should forward test almost identically to the backtest is worth it. I am also exiting with limit sells more often than not. The only time I wouldn't exit without a limit is if the target wasn't reached before the close.

Slippage calculated for the times that it does hit the dissaster stop is 1.5 ticks. Also when exiting at market close if target wasn't reached the slippage is calculated as 1.5. Now for the long forward test in sim 6-12 months. I'll update at the end of every week with results and any slippage figures so that others with the same questions may benefit from this thread.
 
Limit orders: You need to model for price to trade through. Ensure that your software doesn't do a "one-touch" and fill you. This is important for both entries and exits (that will cover your slippage on limit orders).

Market orders: 1 tick of slippage every time. If it's a market entry and a market exit, that's 2 ticks total, per contract.

Stop orders: 1 tick of slippage each time. For live trading, you're best off using a stop-limit that has a limit of 10-15 ticks past your stop price (trust me). Yes, it sounds like stop-market, but that time you don't get filled is the time you really pay for it.

Backtesting is OK to start, but you need to forward test on live data for at least 3 months, better 6 months (I think a year is a lot if you're running the thing daily).

2 cents.
 
You must log in or register to reply here.
Back
Top