The bot’s been plodding along uninterrupted all week as I’ve been busy with work. I’ve checked in and I don’t think there’s been any errors like last week, I’ll post a chart when I can. I’ve had some good hints from Betfair Pro Trader (www.betfairprotrader.co.uk) on how to improve the bots performance. There’s so much for me to do that I’m going to add a projects page to keep track of things. Along with the good advice I’ve received to optimize entry/exit and stake, I’d also like to – add US horse markets to Bot1; run a version of Bot1 on Betdaq for UK horse win markets; move forward with Bot2 as I’ve already put a lot of time into it; set Bot3 going as it does actually work, as in, as a program it works and it’s data collection ability is probably useful with Bot1, just need to refresh myself with it’s function. I don’t think I can do much before Christmas though, too much work on. I’ll keep posting.

Here’s last weeks charts. The first is the UK dogs. Not a bad show with an overall positive trend.

And here’s the Australian chart. 3 stand out points. The two large wins which were lay bets even though the offset lay had already been made. The first lay didn’t show in the “MyBets” page in excel though, so when it came to greening another lay was made. I cant work out why. The large loss was due to an API error, something I’ve experienced before but only on Australian markets. Sometimes it has no effect but the wait time before registering the error is 30 seconds, so if it happens less than 30 seconds before the off there’s no time to get another bet in. I could move last trade time to, say, 35 seconds before off but this requires some code fiddling as the section associated with time is before the market type selection. This is something I’m looking at doing anyway to allow the easy addition of US horses (and possibly others) to the bot. Each error could have gone either way, it just happens to have been 2 wins 1 loss. The chart is a lot more consistent than the UK dogs chart above and a 5 times better return.

BPT asked this question –

Any thoughts on increasing your % profit on turnover?

After some time thinking, I replied –

This made me think. My initial thoughts were to find the profitable trades and target them, cutting out the bad trades through restricting market entry conditions. As this is a relatively simple bot (Bot 1 is basically a WOM calculator with some conditions) I’ve come to realise I’ve probably got as far as I can with it. This is why I’m looking at US horses as a way of increasing turnover rather than increase % profit. I think to make an increase in profit the bot needs to do a lot more work, monitoring movements as well as WOM. This is where I’ve been focusing with Bot 2 but it’s almost like starting the learning process from scratch. Handling large data arrays and working out how and what to calculate, which indicators are useful, what time frames to work in, is bigger than anything I’ve done before in programming. It’s really interesting learning and moving to VB is another challenge. I enjoy programming, working methodically and problem solving, more than trading. Hopefully I’ll make a more profitable bot, one day.

After the UK dogs have finished my bot isn’t trading until 2am when it refreshes markets and starts on Aus horses. During this time there are plenty of US horse races going on. I’ve been reluctant to get involved because my previous experience was that the start times were often wrong. It’s not too much of a problem if a race goes off late, probably less activity and less trading. But if it starts early and suspends I could be caught mid trade. This wasn’t unusual back then. I’ve had an eye on them recently and have seen them delayed(without updating on betfair) but I haven’t noted any early suspensions.

I’m going to restructure my bots flow to allow for easier addition/removal of different event types and then add US horses for late night, early morning trading. I’ll aim to get this implemented within a week.

My bot started to trade these markets on the 18th using settings from a few years ago. After the first couple of days I decided to change the settings, restricting trades to more favourable conditions.

This is where the balance has to be found. If the settings are too loose, the bot will trade many times but may have more losing positions and will likely show a net loss. As there are always going to be some losing trades, if the bot is restricted too much there wont be enough opportunities to make losses back.

The changes were made at around the 27 point and from then the graph has gone in the better direction. This was the Saturday and Sunday markets which may be significantly different to weekday markets in Australia, I don’t know, time will tell. A good first graph though.

Going back to original settings but with some programme flow improvements has increased turnover to nearly double previous week. Settled bets has also increased by over 50%. Happy with a profit but the chart isn’t showing a consistent trend. The initial period, up to around 133, was basically flat, then on to 337 looks good but the tail end is showing a downward trend. As it’s in profit I’m letting it carry on with no adjustments, see what happens. (I’ll update on Aus results later)

No bug to exploit. It happened due to cross-matching and rounding. Someone got matched via cross matching which took a fraction of a penny of my back bet. When my bot cancelled the bet the rounding returned£2, although the real value would have been less. As I still had a part bet on the market when the selection won it returned a rounded £0.01.
I don’t think you could ever be in a position to exploit this as it’s always going to require a cross match that works partly using magic (or a method not fully disclosed by the exchange). Although saying that, you’d think they’d prevent this type of occurrence. Or do they let it happen to suggest there is no manipulation? The mind boggles. Good day.

I see this “risk free” headline a lot but this was actually risk free (only 1p but still). My bot placed a back bet on an Aus horse market and then cancelled (Fill/Kill) after 5 seconds. However I won £0.01 on the market?

This is the place and cancel log entries in Gruss (3719 and 3720)

This is the Cancelled bet in Betfair bet history (2. Archytas)

And this is the Settled bet in Betfair bet history (2. Archytas)

I’m trying to work out if this is a bug I can exploit. Doubt it but just think of the possibility, this time next year…

I’d traded the aus horse racing markets with my previous bot a few years ago, it made sense to give the bot something to do during the night if I was going to leave the computer running. Back then most of the parameters were in cell formula and started with an if statement to determine whether the market was a dog or horse event. Different settings were used depending on the type of event.

When I started up again this year, putting all the program into VBA, I hadn’t allowed for the different event types and the result was I gave up on aus shortly into the project. I’ve now modified the code to allow for different event types.

The settings I’m using for the aus horses are taken from the last couple of months of the old bot. They’re a lot more restrictive of market entry conditions, waiting for more activity than the dogs before starting to trade.

From what I’ve seen, there seems to be a lot more money on the aus horse markets than I remember. I had traded the UK horse win markets with my first ever bot but found the trading too complex for my simple strategy. That’s when I moved to dogs and then aus. I have trialled my scalping bot on the UK horse place markets and depending on aus results, may try aus place markets. First things first, see if I’ve a bank left in the morning.