The pic shows the same algo and chart for the DAX this morning. The chart on the right has Flo turned on - live trading. The one on the left is taking signals from the data stream and posting trades onto that chart.
As you can see, there is no trade on the live chart. This algo enters on Limit orders equal to the close of the previous bar. There was no overlap of price between the two bars and hence no fill was possible. Had I coded Flo to enter the trade on market orders or even stop limit orders, I would have had a fill but at the cost of a tick slippage in probably every trade.
The difference between testing, SIM and live trading is important to be aware of. As part of my process of going live, I do look at all these results. In fact, I run two copies of Flo when I am trading: one live and one that just takes signals as you can see above. This way I can look at what happens every day before I start adding size. Of course will catch all the losing trades and only miss winners. The thing I need to assess is how many winners am I likely to miss and what the impact will be on my P and L. Should I use only market orders? What about slippage? The process of using an algo in fully auto mode requires a lot of sue diligence before increasing to any meaningful size.
I think some of this sentence may be missing: "Of course will catch all the losing trades and only miss winners. (The thing I need to assess is how many winners am I likely to miss and what the impact will be on my P and L. )" as I am not quite sure what will catch all the losing trades…
ReplyDeleteDid you run 2 tests that compared Limit vs. Market orders? Can you share the results, and do you think they may apply to all markets? Thanks again for the posts!
Jedi, "Of course IT will catch all the losing..." Testing results will vary by market, algo, datafeed, distance from broker etc. Sadly no short cuts.
ReplyDeleteHi EL, it is my understanding that Flo runs on 30sec bars and only looks at the bar's close(instead of evaluate per tick). Of course, correct me if that's wrong. But if it is true, it(she?) would miss important timing/volume info that lies between bars, which arguably can be used in deciding entry/exit tactics, such as whether to risk a tick or two in order to fill.
ReplyDeleteAnon 18:29, There are a number of Flos. I run more than one at a time and pick the algo that suits the type of day I think we'll have. Yes, Close only action by Flo. I try and keep it simple and trade the pullbacks in the trend of the chart. There are other Flo algos that trade outside in trades but not usually on the DAX, more ES.
ReplyDelete