Friday, 16 September 2011

Bot botheration


I managed to raise some enthusiasm this week to continue creating automated trading scripts for the horse racing markets using MarketFeeder Pro.

I bought a 6 month subscription for £100 with a view to writing a bot to slowly recoup this (and more if possible). With only 4 months left of my sub I need to get a shift on!

My previous attempt at implementing an automated lay system using MF Pro proved fruitless. Nothing to do with the tool I might add - the rules were easy to implement and I had a couple of good days in test mode - but when I switched it over to live mode one afternoon, it proceeded to find about 5 winners in about 9 races and blew most of my £50 starting bank. 5 winners from 9 - that must be a record?!

Creating a winning bot is no walk in the park...not to me anyway. Anything that has even the slightest amount of complexity to it, usually ends up becoming a major pain in the backside to devise and implement.

The examples on the MF Pro site did come in very handy and gave me a good start, but even implementing a Fill or Kill function takes 3 triggers, and a function to work out how long is left in a race (I'm glad they wrote that function and not me!) takes 5 triggers. So I've got 8 triggers in this cramped little pop up window that I can't collapse and which get in the way of my own triggers that I'm trying to add in there too. I hope they tidy up the UI in the version 7 as it's a bit clunky in places.

Anyway, I reached a point at about 4pm this afternoon where I was ready to let my shiny new bot loose on the win markets (in test mode) and lo and behold, it came up two decent wins and a bunch a scratches in the space of an hour...

"Great, huh?" I said the misses as she passed the office door,

"I'm onto something here" I then exclaimed, smugly.

Through some further examination of the logs though, it soon became clear that I wasn't really on to anything at all, at least not anything intended; it turned out that by programming one condition wrong, the bot was now doing the opposite of what I designed it to!!

So now I've got an anti-bot that's a potential money spinner, while my original idea seems to be a dead in the water


I'm at a wedding tomorrow so what I think I might do is run the original bot and the anti-bot in separate instances of MF Pro (taking care to reduce my refresh rates of course) and let them duke it out while I'm quaffing champagne and devouring canap├ęs* at the reception.

* more like Fanta and sausages on sticks if I know my mate

ugh!

2 comments:

  1. Have you tried gruss, a bit fiddly but good if you have no programming skills but a bit of excel knowledge.

    I've been trying to bot for a while and even when something doesn;t work switching it to do the opposite still seems to cost money :( Luckily it's given me a better idea of how the markets work and how to take advantage. I've a couple of bots now that I'll stick real money on soon, might even blog my exploits if it looks like being a winner

    ReplyDelete
  2. "switching it to do the opposite still seems to cost money"

    Yeah, funny old world, eh? ;)

    I've had a fiddle with Gruss

    http://mets-trading-diary.blogspot.com/2011/07/ive-been-busy-lately-trying-to-work-out.html

    MF Pro is better suited to my needs at this time because of the test mode but I expect I'll revisit Gruss at some point :)

    ReplyDelete

Note: only a member of this blog may post a comment.