I plan on SIM trading for another few weeks to satisfy myself this strategy is profitable. Side-by-side trading will be observed with limit and market orders for entries. All trades will be closed at market. The reason for testing entry limits is slippage control. Currently, front end slippage is 1-2 ticks. In a high volume trading system, that presents an opportunity to capture more profit IF the rest of the system mechanics don't go haywire.
Also, it is clear the performance improves with the addition of the double zero lag EMA filter, so this will be a staple going forward.
Today, I set up side-by-side charts to examine the impact of using limit versus market orders on the front end of 6E, CL and FDAX with the subject auto strategy. The system as currently coded cancels the limit order if not filled after the close of the bar after trigger. (Bars are 2 tick interval.) Not surprisingly, more trades fired with market orders.
And although 6E was more profitable with market orders, the average profit per trade paled in comparison to limit. CL was more profitable with limit, and FDAX lost less money with limit. If this trend continues--and I expect it will--the impact on profitability will be very positive not to mention the decline in risk due to less time in the market.
Tomorrow, I'll trade a couple more instruments. (The strategy isn't curve fitted to any particular market.)
The following user says Thank You to Tundi for this post:
The results are within my pre-conceived idea--limit orders either lose less or win more on fewer trades than market orders. This was probably no surprise and most obvious to you but to me it's sort of an epiphany. I kind of expected this, but witnessing it live, tick by tick is a buzz.
I apologize for not including the chart of 6E market. I inadvertently disabled the strategy and lost the live trade performance data before taking the pic. Suffice it to say, though, the results would have shown a much lower profit and much higher number of trades. (Study the CL attachments for a representative example.)
Last edited by Tundi; December 30th, 2010 at 07:13 PM.
leasing a seat will be your best option to lower exchange costs.. alternative, given the volume you are referring to, look at Advantage Futures ... you can go as low as 10c RT for comms... but as I said, the exchange fees are going to be your issue (as you are aware)
I assume you are using Ninja for your system... and that you will co-locate near or at the exchange... but from what I saw on your charts, you are not doing HFT, you are just always on the market... perhaps i missed something from the charts you posted.
As I said in an early post, this is a pseudo HFT even using market orders with no indicator filter.
With limit orders and the double zero lag EMA (so far, the optimal configuration), I'm not always in the market and quite happy about that. Should the performance continue in the following weeks in correlated and non-correlated markets, then it makes economic sense to scale up the contracts in some sort of logical money management system.
Thanks for the lead to Advantage. $.10/RT? How do they stay in business?
easy... true HFT is execution in terms of volume is easily well over $50K per month... they do millions of transactions per year... and I do mean millions, well over a billion... the infrastructure is costly, but at the same time manageable given their message rate, and it is not all that hard to plan for capacity improvements as volume increases.