Flo is a machine

We are live with Flo, running her on ES and Euro FX futures (6E). I'll show her in operation during the training and after that, if there is interest and I can find a practical technological solution, I'll give a view to all blog readers who want to see her live.

Lot's of challenges to get to the stage we are at. Broadly speaking, testing over a year of tick data using continuous contracts gave a draw down of less than about $2,200 per contract. The last year or so of the 6E had a theoretical nett profit of around $100k and the emini $30k, both per contract.

Flo is not curve fitted in the normal sense. The parameters that may be considered curve fitting are the targets and the stops but these are pretty generic. We are still working on improving exits and will have a later version on test after the week-end.

We test live with real money on one contract with the exit at a logical first scale area. Nothing fancy. The later versions will have more exit logic as well as scaling out.

The biggest challenge is data and connection speed. We use eSignal data for testing and analysis but the orders go to the brokers. There is a disconnect between the prices on the eSignal feed and the broker. We could run the analysis using the brokers' feed but the feeds are not raw enough - snapshots, amalgamations and such like. At this stage, if it makes money we won't worry a lot but further down the line we may look at hosting our server at the exchange to get purer data and faster connectivity. Looks like I may be in Chicago at the end of October and I'll look into it then.

Our current testing is also doing an analysis between the theoretical results on the eSignal feed and the actual trading results generated with real money. We are running this in parallel. I want to see what percentage of the theoretical profit we actually capture.

That's the story so far. We'll expand the number of markets we trade as soon as we get the next version of Flo tested.

Lot's of you are designing automated trading systems. It's not easy but I am now convinced that Kiki's enthusiasm was warranted. Although I still beat Flo by trading more markets at the same time, she wins in the end.

I started posting about Flo because I have found that many new traders look for hard and fast rules for their trading. They are not really discretionary traders in the true sense and perhaps many of them would be more comfortable, and profitable, with a completely rule based methodology that is executed by a computer so that consistency is met, unless, of course, they switch off the PC.

Today is rollover day in the ES and we have had the usual distortions as the spread traders do their thing rollin', rollin', rollin'. The first trade as ES RTH opened was a sell against the upper tail of yesterday. It overshot a bit into a higher volume congestion of 4 June. I traded it down to the POC of the split distribution of yesterday. I then sold it after a rally that tested the highs. The shape of the Profile clued me in.