By better your result I just meant less than 50% chance of losing  I've got a spreadsheet somewhere showing the win/draw/loss of different variations of +1 combat score, 2A and Parry for/against that is quite useful.
Though thinking about it, if you've managed a draw from a losing position you wouldn't be able to parry and so your rules would still work
Thanks for the clarification
. Not still not entirely sure I get what you mean – I'm guessing you mean it's best to take the parry when you have an odds on (>50%) chance of improving your current situation?
I agree this is a good strategy to take, but again, it depends on how you define “improve your current situation” (and of course, it all depends on my selective interpretation of your question
).
As discussed in my previous (and pompously titled) post
A winning assumption, I think the outcome you should optimize first is just straight up whether you are winning or losing  I'll call this Priority I. A secondary concern is to maximise the amount you win by, and minimise the amount you lose by, but only if you have done everything you can to try and win – I'll call this Priority II.
The strategies I gave above actually do
both of these. Here's how:
1) Let's first focus on Priority I.
Imagine a situation where the dice are rolled, and you have already won combat, but the dice scores are such that you could parry if you wanted. Should you? According to satisfying Priority I, the answer is no.
Why? Well, as things stand you have an expected 100% chance of winning combat (you already won, right). There is literally no way you can improve your situation. Forcing the parry, while perhaps letting you win by more, will in almost all circumstances open up the possibility that you might lose. So your expected chance of winning has to be <100% if you take the parry. On average, you would guarantee making your situation worse in terms of our primary concern – the simple business of not losing combat.
Similar reasoning extends to the case when you have already lost. Your current situation has a 100% expected chance of losing, but if you parry there will probably be at least one outcome, no matter how improbable, where you will win. So if you parry your chance of losing is <100%. Your expected outcome is better than your current situation, so you take the parry.
Draws? Well, they could go either way... leave them as they are for now.
This is precisely the reasoning behind the Orlockoptimal strategy, so named as it embodies this optimization perfectly.
2) Let's now ignore Priority I, and just focus on Priority II. We are now gamblers.
Here, we define improving our situation as winning by more, or losing by less. This then becomes a simple question: if we force the parry, do we expect our opponent's score to be, on average, higher or lower than it currently is. Note: the important detail here is the word
score rather than
roll. They might reroll low, but one of their other dice might then step in to become the highest roll, so we have to look at what's going on with our opponent's other dice as well as the one we are rerolling.
This is, broadly speaking, embodied by the Escher strategy. That strategy isn't optimal for the purposes of Priority II, but it's as close as it comes in practical purposes that it makes almost no difference.
For completeness I'll point out that the optimal strategy for Priority II, Escheroptimal, is actually the following generalization of bits of the Goliathoptimal strategy:
 Always parry an opponent's 6

Always parry a drawn an opponent's 5
 Parry a drawn an opponent's 4 if he has no 4's on his other dice, otherwise do not parry
 Never parry an opponent's 3 or lower
I'm not going to go into the exhaustive reasoning behind this (although I'm happy to do so, if anyone is interested) but the principle is simple:
Always parry if your expected outcome, in terms of combat score difference, is higher than your current situation.
I'm guessing this is exactly what you were looking for
@Eldarin?
3) Now let's blend Priority I and Priority II.
First, we try and optimize Priority I when possible. So we follow the Orlockoptimal strategy.
Second, as a fall back, if we have done as much as we can to satisfy Priority I, we optimize Priority II. In practical purposes, this means we try and maximise our combat score difference whenever Priority I shrugs it's shoulders and becomes indifferent – i.e. whenever we have a draw.
This is precisely what the Goliathoptimal strategy goes. How ironic that the Goliath strategy should embody the elegant harmony of brains and brawn, but I digress.
4) As for +1WS, +1I, +1A and their impact?
Well, these (along with charging bonuses, or whatever other combat score modifiers you want to look at) are already built into the curves shown earlier. They just modify where you draw the dashed lines coming up from the xaxis! In my next update I'm going to discuss how to read the curves properly to do this.
Parrys are not included on those curves for one reason – we have to know which strategy we are using before we can measure its impact on combat outcomes.
In summary:
My logic was thus – if I can deduce the optimal parrying strategies, then we can see how those affect the curves in my previous posts. Then, by induction, any other strategies will give worse results. So by just looking at 3 or so parrying strategies we can understand the impact of
all parrying strategies (and there are a lot of them – I've been testing 20 or so, and that's not exhaustive).
Anyway, I hope that actually answered your question
@Eldarin .