https://www.youtube.com/watch?v=5G7CHlge80o
Edgar posts a lot of good stuff here, so in the unlikely event you're not already aware of Edgar TV, then go and check him out, but what I want to do is analyse what he's commented on in the last third or so of the video, namely where in the "cons" section, he comments on tons. Here, Edgar is referring to this chart from Dart Connect:
Now Dart Connect's brilliant, but here's a huge example of where statistics can be either outright misleading, if not telling you the whole picture. Look at that 100+ column, which is what Edgar is thinking looks bad, and something he wants to improve on, and if you look at it at face value, then it does appear bad. But what exactly is it telling you?
Here's where Dart Connect is a bit disappointing, it has a bunch of stats, but doesn't really explain what it does. Here, it just says "all turns of 100 or more". OK, that's fine, but it doesn't look to be calculating it right for one - if we simply count the number of turns where someone scores 100 and compares it to the number of turns full stop, it gives too high a figure for both Edgar, and also for Geert de Vos, who we'll be doing a comparison against. However, if we go for an alternative metric which I thought they might use, and that's to only include turns where it's actually possible for a player to score 100 (i.e. if the marker says Matthew you require 40, it doesn't count), then for both players it gives too low a figure, by more than what it's out using how you think it'd be calculated. Maybe it's me mistranscribing the actual scores, but for it to be the same for both players makes me think there's simply something wrong with how it's worked out.
The main thing is what it's intending to show, and is it a useful statistic? You would think that it would just be intended to show scoring consistency, but a statistic that says someone starting 100-100 is better than someone who starts 60-140, when they both are on 301 after two visits, is surely not that useful. Sure, as a pro you'd want to be hitting at least one big treble a visit, which in theory should correlate to this 100 count, but that's relying on you staying straight all the time and never switching. And here you've got a limitation with Dart Connect, it just records the score with the three darts, it's not going to tell you whether someone scoring 59 has gone 20-20-blocked and 19, or 1-1-blocked and T19. Unless there's something behind the scenes that I can't immediately see (no clue how they score cricket), we're out of luck.
So what's a better metric? The goal of darts is to win the leg, so maybe something like the percentage of legs where a player is under some score remaining after some number of visits. Maybe 120 or under after 12 would be a good measure, so that you see how often you are going to enter the critical fifth visit with an outshot that is only going to require one treble to be on a big double. This could probably be queried if you had a direct connection to the DC database, but for now it'd be something you'd need to work out on your own.
The other big issue is one thing that may be overlooked - certain visits under 100 will cause more visits under 100. Let me give you an example - I start off a leg with two straight tons, you start off with two visits where you block the T20 bed after the second dart and end up with 99. Who's more likely to have a ton visit on turn three, if we assume that we have the same skill level at hitting big trebles?
It shouldn't take a genius to work out that I will, as a result of the previous visits. I'm on 301, can stay on 20's, and will ideally look to get seven marks to leave 161 and be on a finish. You can't stay on 20's, and will try to get seven marks on 19's to look to leave a 170 out. Difference here is that I will get a ton if I only hit one treble - you won't. Having a visit where you need to get off 20's in the early stages of a leg may well see you needing to do the same (rightly or wrongly) later on in order to try to avoid leaving bogey numbers, and through no real fault of your own, and it will really damage your ton percentage.
So is this an issue affecting Edgar? Fortunately we can take a look at this. Here I've grouped every score that Edgar and de Vos put up in the Super Series, and grouped it in blocks of 5, so anything that's 95-99 will be in the same bar, anything 100-104 is as well, etc etc. Let's take a look:
Now I've deliberately left off the y-axis, but as an exercise to the reader, have a guess where the 100-104 group is. If you said it's where that largest blue bar is, then you're correct. de Vos won a game in the Super Series and Edgar didn't, so de Vos has had an extra 24 visits, but that's a huge number of straight tons from de Vos. An absolute mountain. 52 straight tons in 47 legs. By contrast, Edgar has only hit 19, but look at the bars just below those two. Edgar's hit significantly more scores of 95-99 than de Vos has, and enormously more scores of 90-94 - de Vos has only hit three to Edgar's 18.
Is this a problem? I wouldn't necessarily say so. If someone's hitting four straight tons, they're leaving 101 - which isn't really much of a different checkout to 120, you've got 19 points of wiggle room before you're being left with a fundamentally more difficult outshot. If you're on 101 and I'm on 116, then if we both hit a treble first dart then it's just a case of which big number we hit to leave a double of choice. Sure, you might have more choice as to what preferred double you want (if I'm on 114, I can't really set up D16), and there's also you having the option of using the bull last dart in hand on the fourth visit to get just under 100 for the fifth visit in a lot of circumstances, but I think it's fair enough to say that the lack of tons isn't necessarily going to cost Edgar if, as it looks like, it's as a result of a lot of switching.
I don't know if I'll be posting here through the Super Series - I may put in some quick things if anything looks interesting, but for the most part I'll probably just put some each way selections for the outright events on Twitter.
No comments:
Post a Comment