Even though it's been just a couple of weeks since the last newsletter, which was just a short affair, it's time to put finger to keyboard to give you all the latest news about what's happening at UK Horse Racing and to also to pass on a reminder about the Lab Notes, an area of the site which appears to have been forgotten by a number of people of late.
The main news will be, of course, the CSV Data Project which is on-going and is thumping through the data as I type. There's also been a new release of the bot logger which is worthy of consideration and over the next few weeks I will be trying to continue this new theme of investigating staking plans.
I hope that this edition is of interest to you.
Over the past week or two three versions of the Bot Logger have been released. The first version of these three, 1.25, added a new feature to the Portfolios sheet.
This feature was added when it we wanted to investigate what the effect of having a large number of portfolios running on one bot and we wanted to see how smooth each of the portfolios' profit/loss line was. So for v1.25 a simple calculation was added to the Summary sheet of the logging spreadsheet (below).
Figure 1. Columns H & I showing a new bank calculation feature.
The Bank figure, in column H, works out what would happen if that portfolio, and only that portfolio, were backed to 1% stakes with a 100 point bank.
We see in the above screenshot that the first portolio is the UKHR-1 Maiden Stakes which had had 221 runs, had 84 winners and made 61 points to level stakes. However, with a 1% proportional stake the bank would have grown 77% to 177.93.
The 1% stake is, by the way, configurable by changing the value on the Summary sheet of the logging spreadsheet. Yes, you can see what damage a 2%, or even 3%, stake will cause to the smoothness of your bank!
All well and good, but this alone doesn't tell us how volatile this portfolio's history would have been and this is where Column I comes in.
This figure shows us, in percentage terms, what perentage of the bank went down to in its history of these 211 runs. In other words, the largest drop from one point to any subsequent low point was 10.2%. This indicates that the UKHR-1 Maidens is very smooth.
Compare the UKHR-1 Maiden with the Frm - Alarm plus portfolio (row 15 in the above screenshot). The same Profit (61 pts) was achieved but at a lower strike rate and with a better ROI. But look at what happened to get there; the bank would have fluctuated a lot more as the deepest point the bank dropped to was 73.2%. This, clearly, is a rather volatile portfolio and I would be hesitant to include it with anything else.
When it comes to the flat season the Cls Flat is going to be employed; a good strike rate, an excellent ROI and, most importantly, with the lowest dip at 91.3% it looks as smooth as we can expect. That portfolio, for sure, will be employed somewhere here.
The first signs of feedback was enouraging so this functionality was added to each of the five race types (Flat, All Weather, Hurdle, Chase and National Hunt Flat) so that the same thing could be examined in particular race types in case one offered a more stable environment than the others. The same was further added to the Stakes and Handicap sheets too.
All the above is all very well and good, but it does stop one step short of the mark. It doesn't take too long before the question is asked, "But what happens if I wish to have more than portfolio running? How would they react together?"
Good questions both. So, in version 1.27 a calculator was added. Clicking on the Bank Analysis button on the front sheet a dialog box will appear.
Figure 2. Bank Analysis.
Thankfully, it's not as frightening as it may appear particularly as I specialise in dialog box styles of the 1990s.
In the left hand list all of the portfolios in the log file are listed. The button with the double arrow places the portfolios into the box in the upper right. These are the portfolios which are going to be processed together as if these were in a bot of their own.
There is a Stake% box to the bottom of this which enables the user to change the percentage stake of any of the portfolios if so desired.
Click on the Calculate button and then the runs from these portfolios, in this case three of them, are gathered, sorted into time order and then analysed. This, then, ought to give the user a good idea of what would have happened.
So, in this example we're looking at three flat portfolios; RAdj Flat, Cls Flat and Spd Flat. Each of these three have been given stakes of 1% and the net result is that after 398 runs the bank showed a profit of 698 points and with only a small drop at any one time - just the largest drop of 21%.
These three are certainly going to go into a summer portfolio here for sure.
...what if I am not a bot user?
Well all is not lost. It's still possible to find interesting selections from, for example, the log files such as the ones above and then look for the definition of the particular portfolios within the UKHR Bot Portfolio Files page.
Then it would be rather straightforward for those who wish to do these selections manually to do so.
The Results Archive Project is well under way.
For those who have just 'tuned in' to the Newsletter the idea is to tidy up all of the CSV files (going back to January 2005) so that the format of each of the files is identical and then to back populate any data which we can.
The format of the data files has been sorted out some time ago and now we're going through each of the two thousand odd CSV files filling in statistical data such as the Trainer form, the Trainer/Jockey combinations and so on.
Care is taken to populate the data so that the data would be accurate for that day. This means that for a CSV dated some time in, say, March 2005 then we're looking at the historical data between the day before that CSV file was generated and five years before that; from March 2000.
This is taking quite some time with over 350 CSVs for each year to process but one things the computers can do well is to churn and churn continuously. At the moment it's taking about a day and a half to back populate the missing data for each year, so it will be a few more days until we've got the full lot between 2005 and the end of 2010.
If everything goes to schedule then the results for 2007 should be ready sometime on the 25th of March, 2008's data should be available sometime on the 26th and then we ought to follow with 2009's on the 28th of March.
The results for 2010 will be ready a few days later, perhaps closer to the weekend of the 2nd and 3rd of April.
If anyone isn't familiar with what the CSV files actually look like then a sample file of the 22nd April, 2006 CSV can be downloaded from here. A full key to what each of the fields means has been produced and published within the Archives pages and will continue to be updated whenever appropriate.
Quite simply the purpose of this exercise is to gather years worth of data together so that it can be analysed.
It is designed for people to import into databases of their own creation or into spreadsheets and then to be looked at in order to find new angles and systems and as well to test out staking plans. Having the data in a handy format means that anyone can look at the data and quickly check anything and to explore as much as they want.
It also makes it easier to answer any questions such as "How good are your ratings?" All we need to do is to point to the results and say that the orginal ratings are there and all we've done to them is to add the results and the quality of the ratings are free and open to anyone to inspect.
But the real purpose is that this data is going to be there for people to investigate and to find new angles. Which brings us neatly onto the next point, the Archive Working Group
As expected, this project has caused a bit of excitement and quite a number of people have expressed an interest in the project.
In fact, so much so, that a proposal has been raised by some members that they would like to form some form of Working Group to analyse the data, each in their own way, and to pool any findings and to share ideas for research.
I am starting to collect names of those who would be willing to participate. There's only two real conditions, I suppose. The first is that you're willing for me to pass your eMail address and name around the group and, secondly, that you're willing to actively do some research and to put forward findings and ideas.
I won't be running the group but all I will be doing will be to gather the details of the initial group and, so, if you're interested and willing to put some work in and share what you find then please get in touch. I will, of course, be involved as a member but I won't be running it. Hopefully it will just run itself.
My initial interest in the Working Group will be to go through the portfolios which I have and then to see if the filtering on them is correct or they could be improved in any way. My second item of interest is that once that I have confirmed what I am doing is right (or having improved on what I have) is that I want to explore the world of staking.
In my view this is the only part of the whole jigsaw which we haven't fully sorted out. Lots of internet sites have been filled up with selections but little has been said about staking. I hope to do lots of research into this area to maximise profits.
A copy of my logs (until the 24th of March. 2011) can be downloaded from here (2.3 MB) .
These logs are created with the latest version the Bot Logging Spreadsheet and even though the stakes are removed other useful functions, such as the Bank Analyser, can be employed to see how various portfolios can interact with other.
All this talk of CSV Archives under production and 18 months of log files doesn't truly give the big picture.
So let's not forget the Lab Notes which contain over sixty different spreadsheets and reports all of which help the user to choose strategies. For example, in there we have spreadsheets containing data going back for the years 2005-2009 inclusive. We have reports on how the various systems and rated horses have done over the last few years - all vital stuff which is not to be forgotten.
Even when the CSV Project is updated and everything is placed in databases the Lab Notes are still going be a useful place to look through to get ideas for initial areas of exploration.
In fact, I don't forsee this section of the site wilting when the CSV Results are completed. One of the tools which I have been writing here is to use the finalised CSV files and to make reports and Excel files in much the same way that Holmes has done for us.
So, do expect more and more items filling up the Lab Notes in months to come.
First of all, the heartiest of congratulations to Herby for his victory in the Cheltenham Festival competition with a high priced winner in the last race of the competion ending up with 28 points to the good.
Herby's score was shown to be even more extraordinary given the results at the Festival this year with most of the competition entrants making an unusual paper loss.
Well done, Herby.
That's all for now for this newsletter. So, until the next time, see you in the short queue.
The UK Horse Racing Team
PRIVACY NOTICE AND NO SPAM POLICY:
We at UK Horse Racing take your privacy seriously. We have a strong policy of not handing over our address list to anyone at anytime. We suppose that if the police come bursting through the door with a warrant and a brace of half-starved alsations then we may have a sudden and inexplicable change of heart. But, understand that otherwise we won't give away, sell, loan or whatever to anyone your e-Mail address. In addition you have received this message because your address was placed on our mailing list. If you didn't do this then some scallywag registered you on your behalf, so please don't blame us and call us rude names. We have set up this mailing list as opt-in only; we didn't harvest your eMail address from anywhere; we didn't take it from the members lists, the forum, any competitions nor did we beg, steal nor borrow it. If you don't wish to receive any more communications from us then please return this message back to us and we will remove you from the database and you won't be bothered ever again.