With regard to bank sizes this may be of assistance. Whilst I don’t have Ray’s historic data with regard to the actual winning strike rate of his selections a historic statistical analysis of the past 5 years in terms of races won by horses over and above 16/1 may help and shed some light.
So taking 8+ runner races in Flat handicaps, Flat maidens, NH HC Chases and Hurdles, NH hurdle Maidens and NHF races equates to 36,243 races. I may have missed out a few race categories, but with such a large sample size it won’t really matter. Additionally I know that races with odds on shots are eliminated and that winners may end up below 16/1 at SP, but in theory with such large sample sizes all these factors will equal out. From these 36,243 races 3560 were won by horses over and above 16/1 – a 9.82% race Strike rate. So broadly speaking 1 race in 10 of all qualifying races at all UK courses races is won by a horse at 16/1+ at SP. Insofar as individual selections are concerned the winning strike rate is 2.10% or one winner per 48 selections, illustrating the fact that on average there are nearly 5 x 16/1 selections per qualifying race e.g. larger fields. Taking the worse case scenario of a 2.10% individual selection strike rate by using this http://greyhorsebot.co.uk/strikerate.asp we identify a 383 longest losing run over a 5000 bet sequence.
From my experience a total elimination of a bank occurs not because of a unique single long losing run, but a series of poor results with a winner or two sandwiched between two or three poor losing runs, so to safeguard any bank you should factor this in. I appreciate that Ray is selective with regard to which races race are qualifying races so the above will be an absolute worse case scenario and a true picture can only be calculated with the actual historic strike rate of Ray’s selections. Should the actual historic strike rate of winners per selections equates to say 4.2% which is double the above example then we can expect a longest losing run over 5000 selections to reduce to 189 bets. By knowing the historic strike rate of winners over selections the above link can be used to work out the longest losing run and subsequently the bank size you may need.