e-mini slippage

[Sircher]

I developed a strategy and it seems great on backtesting with no slippage. Probably could handle 1 tick slippage on both sides, but when I compare the backtest to simulated trading of the strategy there are 2 tick, 1 tick etc. difference in orders ( to the negative of course) my concern is that in live trading the slippage would be this bad, one or two ticks, none of the trades were the same on both charts sim and backtest. With the volume of trading in emini, why should there be such slippage. I am only intending to trade one contract right now. Is internet speed a factor?

I was thinking that trading with limit orders might work, but I cannot figure out how to get the limit order to generate. In the Conditions Set 1 I have a criteria, but in the Action field I put what I thought was the terminology to enter a short limit order at the close of the last bar price plus 1 tick on the Ask. Nothing shows up on backtesting and it works fine with market orders. Very frustrating.

[NinjaTrader_AdamP]

Sircher,

There are known discrepencies between live trading (even simulated) and backtesting. For more information please see the following link.

http://www.ninjatrader.com/support/helpGuides/nt7/index.html?discrepancies_real_time_vs_bac.htm

Please let me know if I may assist further.

[nicknamed]

Sircher, use same data provider for backtest and live data, test and compare results on that source of data, I recommend IQ Feed, based on my backtest and live trading on IQFeed data, average slippage is around 0,75 tick during Regular Trading Hours (RTH) session.

[Sircher]

slippage

Thanks for the reply. I am using CQG. Can there be that much difference with feeds? My system can probably handle the .75 slippage. Thanks for replying!!!

[nicknamed]

yes, there is striking difference between the feeds even on these centralized markets. this pring when we turned our strategy from ETF to futures I did thurough analysis, went on many demo accounts of different feed providers and found there are huge differences I did not expect, so I decided to trade on same datafeed I tested my strategies on, so I am on IQFeed which is the only working feed, CQG has image of ultra high quality datafeed but according to my analysis is not much better than Interactive Brokers or Trading Technologies (say Velocity futures where I am). So I am testing all strategies on IQFeed data (which accroding to my test are the only data that go directly from exchange, all other data are biased by a broker's own customers - I do not understand that why, but I take it as a result of the analysis) and also trading live on IQFeed while ignoring datafeed from Trading technologies when trading on Velocity futures. TT has delayed data and is one of the worst datafeed services around, but this way I am attaching great IQ datafeed with good broker and I am satisfied.That average slippage is the result of my analysis from spring and truly, since then when trading live (my system goes for the open at next bar) I have rarely slippage. Usually on low liquid markets and times. During standard hours there is no slippage (I trade ZN, ZF, ZB, ES). You can try demo from IQ Feed to compare backtest results of your system. always test with commission and slippage of 1 tick for securities mentioned and 2-3 with other relatively highly liquid futures (CL, TF, YG,YI, GE,DX, 6B, 6C, 6E). write me if you need more tips/info.
N.

[nicknamed]

May I ask you what do you mean that your system may be able to handle 0,75 tick slippage? Is it that with slippage of 2 it will turn negative? If so, forget such system, I am not running a system live if it got average trade under 2-3 ticks...as that is a suicide. My current system has like 5-30 ticks on different assets so I am fine and normally to stress test the system I use slippage of 3 ticks (even if I know average is under 1), I overshoot commission and play with monte carlo analysis and never run a system live if it goes to surplus behind 10 % number of occurences with 5 % best winning trades cutoff and 500 simulations. Example: my system with 5 % best trades cutoff w/ 500 simulations must - including 2-3 ticks slippage above average slippage (you never know what will happen in reality) - turn positive under 10 % of all simulations, I usually have 3-7 %, but sometimess I have also 10+ and I never run system with such settings, better choose more conservative settings with maybe lower profit, but also significantly lower risk and I am really risk averse...

[Sircher]

My system is trading for more than one or two ticks. Usually 3-5 points. I don't think it can survive 3 tick slippage on each side that seems excessive to expect a system to require that, but I could be wrong. Maybe you mean total slippage round turn, if so, it can handle 4 ticks total on a completed trade, buy and then sell.

The Monte Carlo statistics that you mentioned are confusing. I guess I don't understand them. When I run a Monte Carlo simutation with 500 simulations cutting out the upper 5% and lower 5% it comes out positive from 1%. Probably doing something wrong. Thanks for your response. At some point I will have to just trade. I am using CQG since Ninjatrader uses Kinetick and you say it is the same as IQFeed it may be a better way to go.


Thanks

[nicknamed]

4 ticks should be fine with good data and liquid market.
What markets do you trade?
Do not cut the lower 5 %, you want to see if the system can be any stable when loosing some profitable trades while trading with losses.
Kinetick is just another provider, and every provider has differences. Read huge discussion at Big Mike trading about data differences, their result is the same, IQFeed while I think they did not compare to CQG, I did and CQG was not any better than mirus or trading technologies. Just different tick data and you could see that from charts. Try not backtest but simulation at CQG, try different demo account with different broker that uses CQG if you have same results on CQG data on demo, if not, something is wrong and you should test more before going live.

[Sircher]

Just e-mini.