Ok, brand new feature now!

Deal-Eye (as I've coined it) is a fairly complicated Excel Stats Calculator which I have devised to try and determine the offers that the player should receive based on the board (what is the mathematical minimum they should accept, as a theory).

Firstly, it works out the "Fair Deal" of all remaining final 2 combinations and divides by the number of pairs left to give what I've loosely termed Expected Value. But - and this is the clever bit - the Expected Value decreases depending on volatility/ instability of board. So, a board with 4 blues/ 250k will compute a much lower estimate in relation to the mean than a board with 5 reds, even if the average is similar. The fulcrum that volatility is based around is the "Median Pair Value", which at the start would be the value of the 116th-highest pair.

So, basically I'm going to put up a graphic most days (when I can) showing what the calculator recommends and what it might tell us about how the banker views the player. Feel free to leave comments.

When the player Deals, it will be an analysis of that Deal (except at 2-box)
When they play to the end, I'll flag up an interesting decision point.

NOTE: They'll probably mostly be day after broadcast as I'm looking to add a pic from 4OD.

(P.S if you want the Excel file, I'll happily send it to you on certain conditions)

Ok, here's the first one:



So, a pretty cast iron DEAL recommended today.
From this offer, I would infer that the banker views this player as semi-aggressive.

Ok, Today's verdict:



A NO DEAL is today's strong recommendation, which implies a cautious decision. Boards with a blue instead of £75k have seen similar offers where players have reluctantly dealt!

(I should state actually that this page is purely about interpreting Deal-Eye's results, not necessarily my view, although there is often some correlation)

Interesting stuff!

What would Deal-eye done at 5-box today?

I think the benchmark figure was at £52k, but I don't think Tiffany would've been offered as much as the 55k that was quoted...

Today's game:



This is an example of the calculator taking into account volatility of board, so a solid DEAL today, despite the offer barely scraping 40% of the mean. All we can really infer from this is it was a sensible board reader call.

Yesterday's game then:



A NO DEAL is the verdict, a slightly cautious decision but a justifiable one with the split the wrong way.

5-box was interesting - calculator said a tad over £11k, and the banker said £10,600, but that was a marginal call of NO DEAL (from Deal-Eye's point of view!)

Today's game then:



An overwhelming NO DEAL, and I'll explain why in a minute.

Now, as an analyst, these are the kind of boards I love, as they create a lot of interesting debate. It is the classic Board Reader vs Targetist board in my opinion, and both sides have a fair case for their decisions:-

The targetist sees £30k utility-wise as a significantly 'useful' sum of money and the utility drop to a blue as being far steeper and far outweighing any reward potential because their pain threshold has been reached. Going away with a blue is just unthinkable to them.

The board reader is prepared to take risks perhaps beyond utility calculations due to strength of board. They are more inclined to gamble on stable boards, and I've tried to reflect through my calculator that mathematically speaking, you should be. (62% of the mean is today's benchmark)

However, it's not surprising that people are split as to whether to take the £30k. Mathematically speaking, it's a fairly poor offer, but I think this illustrates how difficult it is to play the game purely on maths, because for a lot of people outside factors (work/home etc.) have to be a consideration, especially in the current economic climate.

The higher the stakes, the more likely it is that declining marginal utility becomes a clear factor. Utility functions might well be better approximated by a logarithm than a square root. (The latter being the central assumption behind the FD of course - I did try and fail to create a log-based model back in 2005 though...!)

Finishes like today's, of course, are why players tend to be targetists, especially as the pressure not to be has rather reduced these days as Noel has gone from being consistently interventionist to inconsistently interventionist.

You've forgotten one other challenger...

The wakeyist.

Yes - but you know their decision already

Fascinating board anyway, and the Banker these days knows that he doesn't need to play the percentages for many players on those boards, which has secured him several huge wins... Not this time though!

If I had designed that calculator, here's what I'd do with the graphic design:

I would use white for everything except for the title card, contestant/date cell, cells "Probabilities:", "Finishes:", "Offer:", "Benchmark/Verdict", and even there, I suggest you use much lighter, much less intrusive colors. Note that light is not bright, so the green in "Finishes:" is intrusive, bleach out the color more towards the white end (use the color editor in Excel and bring the slider up towards the white end)

The board could just be shown with normal cells which I think would be cleaner- the actual board uses a Eurostile font, which I think is a default font. Just use a solid gray if the paddles have vanished, the lines are distracting. Not sure what to do with the squished contestant's photo.

Essentially, simplicity is better. Use fewer colors and emphasis styles which have the tendency to clash, and you're in good shape.

---

Enough of that.

I love this calculator, thanks so much for designing it. Ever since Glenn's game from 30th November, I've been maddened every time someone turns their back on a board that promises to have so much going for it. Franca had the dream board among 20K-100K and 20K-250K lurking among it, and the probability of kicking all 3 as listed was 10%. Now of course, I got my butt spanked with the Banker, as she ended up with the gigantic offer, and I ended up with the quid. I still think the statistical way is the best way to play the game.

The board readers and I thank you for this gadget.

Firstly, thanks Yabbaguy for your comments - I welcome any feedback from you guys about my new 'device'.

I was actually thinking the other day aesthetically it could look better on here - I think the trouble is being the designer I was used to what each colour scheme was telling me, and I know where to look, unlike of course a first time viewer who is going to want it to be easy on the eye. It's the balance between not making it look too dull, but making it understandable. But yes, some of the colours need paling out more certainly.

On another note, thanks for saying you like it!

You know we're all avid fans of your stats...I need to have another look at your site (if it's still up )

Yes, I've been a bit busy so haven't updated the site for a while now, probably something to do when the Summer Break comes around!

Today's game makes interesting reading:



Experimenting with a different colour scheme you'll notice, but I like to keep the 'bubble' display graphics for the amounts really.

So, exciting - the first MARGINAL NO DEAL of the week!
(This is when the estimate is £1,000 or less either side of the actual offer, and is highlighted orange)

For games like today's at 8-box, shouldn't the probabilty statistics be changed from top 3 to top 2? Top 3 is rather irrelevant.

Good point! The numbers change automatically dependent on cell conditions, so I haven't given much thought to it, but duly noted!

Today's all important verdict then!



A fairly overwhelming NO DEAL then, which is not that surprising really - I was thinking £26k, calculator reckons a slight bit more at £28k. However, on the flip side, the deal was at least understandable with a 30% chance of disaster and virtually a 60% chance of a decreased offer...