Transaction Costs / Exchange Fees

[ensemble]

I ran my system on paper in May. The system made a total of 718 trades on 4250 contracts. The activity increased in the last week of the month after I made a change to the strategy. Attached is the paper statement (ignore the discretionary ES trades).

The main problem I found was that the commissions (sim) totaled $4479, or 25% of the MTM proceeds.

I am using the IB Adaptive Algo on "normal" priority with the limit price = ask + (ask - mid). In most cases, the order fills partially at the mid price, and the remainder at/above the ask. I have found through backtesting on tick data that missing a trade outweighs the cost of a "liquidity taking" order, but the exchange fees clearly add up.

I am curious if these costs look reasonable to other traders here with similar volumes, or if the simulated fills are far off base.

 

[cruisecontrol]

Spending 25% on fees is reasonable for a high turnover strategy. I would be much more concerned about the simulated fills, unless you know exactly how it works how can you trust it? I would always recommend writing your own tick backtester to be sure.

 

[qlai]

I would always recommend writing your own tick backtester to be sure.

I wouldn't even bother for options ... Just have to forward test it I guess.

 

[MattZ]

Spending 25% on fees is reasonable for a high turnover strategy. I would be much more concerned about the simulated fills, unless you know exactly how it works how can you trust it? I would always recommend writing your own tick backtester to be sure.

To send paper trading statements is brilliant. Wish my FCMs marketed it this way.
100% right, paper trading has zero slippage and you are always filled on the way in and out. Zero latency issues.

 

[nooby_mcnoob]

To send paper trading statements is brilliant. Wish my FCMs marketed it this way.
100% right, paper trading has zero slippage and you are always filled on the way in and out. Zero latency issues.

For what is worth I don't see paper trading fills on options sometimes. With IB

 

[ensemble]

For what is worth I don't see paper trading fills on options sometimes. With IB

It seems fairly realistic. Below is the SPY 5/31 280.5 put on 5/30. My limit price was 1.87 and the IB Adaptive Algo delayed the order 15 seconds before filling at 1.8 and 1.78

 

[cruisecontrol]

I wouldn't even bother for options ... Just have to forward test it I guess.

I did not realize this was options.. you're probably right, backtesting might be impossible here because of illiquidity.

 

[nooby_mcnoob]

It seems fairly realistic. Below is the SPY 5/31 280.5 put on 5/30. My limit price was 1.87 and the IB Adaptive Algo delayed the order 15 seconds before filling at 1.8 and 1.78
View attachment 203402

Sounds to me like you should put some bucks on it

 

[d08]

You cannot use paper trade fills to estimate real fills. The market will react to you and it's always adverse.

 

[jonahern]

I ran my system on paper in May. The system made a total of 718 trades on 4250 contracts. The activity increased in the last week of the month after I made a change to the strategy. Attached is the paper statement (ignore the discretionary ES trades).

The main problem I found was that the commissions (sim) totaled $4479, or 25% of the MTM proceeds.

I am using the IB Adaptive Algo on "normal" priority with the limit price = ask + (ask - mid). In most cases, the order fills partially at the mid price, and the remainder at/above the ask. I have found through backtesting on tick data that missing a trade outweighs the cost of a "liquidity taking" order, but the exchange fees clearly add up.

I am curious if these costs look reasonable to other traders here with similar volumes, or if the simulated fills are far off base.

this is apples to apples I guess b/c I use IB with same algo's and have a mix of taking/providing liquidity and a mix of algo "priority" settings... I am seeing comm of
-1.281, -1.236, -1.223 per rt for each of last 3 months.