Last week’s charts both end positive. The UK dogs is still going good, the last week in loss was back at the beginning of January.

The Aus markets traded to a profit since being flat (excluding errors) for a month. I’d previously set the Aus stake to £2 to avoid the turnover charge and was thinking that using the same stake may make my bot stand out. Or I may be paranoid. Well I changed it to a percent of bank earlier in the week, so the stake is different most races, if only by a penny or two. I hoped this would provide some cover from bot stalkers/baiters. The proof isn’t here yet but it hasn’t had an immediate negative effect.

Not a bad week. The UK chart is very nice, a good steady trend. The Aus has some errors but ends positive. The general trend gives hope that the move to VPS may improve things. The UK and Aus markets use the same greening code and the errors are only present in the Aus ones so I can only assume the problem is either with the delay or Aus API.

Starting with the UK dogs, a poor result. The two large losses early in the week were both multiple losing trades on one selection in each event. The rest of the week had a general downward trend. I wondered if Cheltenham had any effect. I know it impacts on all the horse racing as people get involved with other venues whilst trading/betting the main races. Maybe some dabble on the dogs as well.

The Aus markets returned well although errors persist. At around the 125 point Gruss was updated with a fix regarding the error related to the bets manager. I was hoping this would fix my errors but the second to last market had an error similar to what I’ve experienced previously, with a lay placed in the last seconds before off. I can’t work it out from the data I have.

Another issue I had came on Saturday afternoon when I decided to update my PC. Saturday afternoons are usually quiet, probably because most people focus on the horses or football. As I’m on Windows 7, I don’t get update notifications, only the GET WINDOWS TEN FOR FREE message every day. So I have to go into the control panel to check for updates and there was about 50 pending. After an hour of updating it failed and, if you’ve ever had this happen you’ll know, it then needed to be left overnight to revert back. This meant I missed Saturday evening and Sunday afternoon on the dogs and all Aus races on Sunday. This is an occasion when a VPS would have been brilliant. More on the VPS project later.

I came home the other day and found my bot had stopped due to an overflow error.

After each market is traded the “My Bets” page data is added to a log sheet along with bet attempt data. Every week or so I save this sheet and clear it in the bot.

When running, the code finds the next empty row to paste to but that row number value is an integer data type. This has a limit of 32767. As can be seen, I’d left it too long and it reached it’s limit.

The Aus trading has experienced some errors lately (see chart, they’re the sharp jumps up and down) which I put down to my underwhelming internet connection. However, I was reading some posts over at Gruss and there might be another explanation. This post talks about an issue with the Aus API where bets are not updated in real time (if you log into the Gruss forum there’s an acknowledgement of this error in the Bug Reports section). The errors that I have looked at occur close to the off which would make sense if the bet status is delayed whilst my bot is greening; it’s calculating on incorrect data. I could stop trading further out but I don’t really want to as some markets only start getting active in the last few minutes. I’ll employ the ‘wait and see’ approach to this problem.

Here are the charts for the previous 4 weeks.

The UK dogs performed below its recent average of 0.061% but with no errors being recorded, this is ok.

The Aus markets returned nearly double the recent average but the errors are obvious. Interestingly, if the errors are removed the end result isn’t much different. This could be an example of the random nature of the errors and has a nil impact over the long term. Still, I’d rather the errors didn’t occur.

An overall good week. UK dogs made a good return with 0.084%. Looking at the chart, the sharp drop early in the results was on one runner in a very active market – 46 settled bets in less than 3 minutes. Of those, three backs were placed around 5.8, followed by a fast drift against which my bot chased and partially matched at 6.2 and 7.5, then finished on 8.

The Aus chart looks good but holds a warning. The two big wins were due to errors. The first was a Lay placed when there shouldn’t have been. The log shows a Lay placed with no Fill/Kill or Tick Offset. My bot only places these during greening so I can only think a matched bet was showing as unmatched and then changed afterwards, as my records show all as matched.

The second was a Lay placed 15 seconds after the start time. This should not be possible as my bot jumps to next race at -1 second. I can only assume that the connection was hanging and this had an effect.

I think it’s safe to say that after a number of errors recently that have left me with a profit, I will soon experience a loss from an error. As they tend to be Lay errors, I assume the loss will be large.