Rolling Thunder Forums

The purpose of this post is to explain...
- what is and is not capturable regarding blocks
- how BBDB will improve om BBM in this regard
- for you to point out anything I say that doesn't make sense, or things I haven't considered
- for me to write my thoughts down so I can clear some space in my head

What information are we trying to capture from a block?
Here is the aspirational list in the order that they are covered below. I don't think all are entirely possible.
- Who is doing the block
- Block dice rolled
- Block dice chosen
- Type of block (frenzy, multiple block, ball & chain, normal block etc.) (new for BBDB)
- The odds of a turnover occurring (new for BBDB)
- The outcome of the block (BBM offers a tick, cross or blank, BBDB will offer a longer list of outcomes)
- Who is being blocked (new for BBDB)
- The odds of the block being a success (as defined by the other player being on his backside without it being a turnover, or the Ball is Stripped) (new for BBDB)
- Who gets to choose the block dice, eg is it 2 dice for or 2 dice against (new for BBDB)
- Was the block re-rolled / is this block a re-roll (new for BBDB) (this is comparatively easy and won't be covered in this thread)
- Anything else?

OK, so all I have to work on is what the log file outputs. In the example of player X blocking player Y, this consists of...

1) [Optional] player X uses skill Z
2) Player X rolls block dice
3) [Attacker Down] - [Defender Stumbles] (for example)
4) [Optional] player X uses skill Z (for example, Block)
5) Player X chooses [Defender Stumbles]
6) [Optional] Player Y uses skill Z (for example, Dodge)
After that there's a number of things that could happen. We will call these post-block rolls and will cover them in a later post.

Easy ones first... lines 2 and 5 tell us who is doing the blocking, what block dice are rolled and which block dice is chosen.
Line 1 reveals if a special sort of block is being used, if it is not there then it is a normal block. So there's 4 things captured already and with storing the sort of block we have already gained some functionality over BBM.

Next up, the chance of a turnover. I know who player X is, and I have his skills stored elsewhere. Therefore I think this is straightforward - it's the chance of rolling all Attacker Downs or Both Downs, or if the player has block/wrestle, it's the chance of rolling all Attacker Downs. Does this sound correct? I don't think any of the defender's skills can change this, unless player Y has wrestle and player X has neither block or wrestle, in which case if the defender chooses to use it then Both Down is not a turnover. However I think it is reasonable to ignore this case - the attacker has lost the block and the turnover is entirely in the hands of the other player.

Now, the outcome of a block. This is derivable by the post-block rolls. I will come to this in a later post.

Next up, things get tricky. Who is being blocked? If you look at my log file extract above, you can see we don't always know who is being blocked... we only do if any of the following happen.
- They use a block-related skill (dodge/fend/any others?)
- They have to take a post-block roll (eg an armour roll)
- The blocker has frenzy, first block is a push (doesn't reveal who is being blocked) but the second reveals who is pushed, we can then put that player against the first block.
- Any other scenarios you can think of?

OK, so we can log the blocked player where we can, but we aren't always going to be able to. So I can't, for example, report on how many times a player has been blocked.

This has implications for calculating the odds of the block being a success. Specifically, if the defender has dodge and the attacker does not have tackle, this modifies the chance of success, but if we don't know if the defender has dodge then we can't take this into account. There are some things we can do to try and get around this.
- If the attacker has tackle, it does't matter if the defender has dodge.
- Look at the number of players on the defending team with dodge. Use this in the odds calculation, eg if 50% of the team have dodge, make the odds be half way between (odds if player has dodge) and (odds if player does not have dodge).
- Go a step further and how many players who are still on the pitch have dodge and use that in the calculation.
Points 2 and 3 are quite complicated and will have to wait for a later release. I may implement a simpler race-based solution for now, eg if team is amazon, then they have dodge, if team is elf, 50% chance of dodge, else defender doesn't have dodge.

Now it's who gets to choose the block dice, eg is it 2 dice for or 2 dice against? Curiously the log doesn't tell us this. So it'll be a case of deriving it where possible from supporting information. For example if there's a red skull and a defender down but the red skull is chosen, we can assume it was a 2d against block (exception - if the block happened as time ran out and the game picked one at random - I think I can capture this scenario??). Or if a ST3 player (without block) is attacking a ST5 player, maybe that's 2D against?

For all these cases where we can only make educated guesses about the information, the fact that we are guessing it will be stored as well.

=============================

I think that's everything? Your responses will be greatly appreciated!

***Post reserved for discussing post-block rolls.***

I'm sorry if this is a half hearted answer, but I'm very busy and tired today;

But did you give an option of what outcome was chosen? After a whoopin' last night by La I'd be interested also in some kind of 'overall' success Vs what you rolled - ie La rolled a double skull, so re-rolled into a skull and defender down. So, in the land of 'Chance' that was unlucky dice; but in the land of 'Nuffle' that was successful (and jammy). Does that make sense?

I suppose what you'd really want to analyse each block would be a way of looking at the situation where the blcok was taking place, and 'scoring' the results that would give the best outcome for that situation - but I guess you'd also need to know what was going on the coaches head too, which makes that impossible. (I.e. if you're blocking someone on the touchline with block and dodge then the 'positive' outcome chance is much higher than if they weren't on the sidelines; because defender stumbles and the two pushes now become more relevant).

Im probably blabbering. Sorry.

I think it's going to be very hard to determine a successful block due to the number of potentially desirable outcomes. The obvious example I can think of is surfing someone and therefore choosing to push them into a surfable position rather than knocking them down. You'll be able to track if a player gets surfed but what if it's a chain push to surf someone else?

I can't see any way of defining successful block beyond "didn't cause a turnover" and even that's situational (I take a 1d block which comes up skull against a skilless player to avoid a cloberring from a ClawPOMB monster, or a both down result to sack a ball carrier).

Yes Rav, I think you're right.

Very brief answer/suggestion as very busy, sorry.

The UKBBL uploader shows a table of “Best Dice in Block Totals”

So it shows, for example:
Player 1 (me) player 2 (daigaro ogami)
Skulls 1 42
POWS 19 0

I quite like that.

Direct quote from last nights game.

Honest.

The 'Who's being blocked bit' - If they get twonked over then Armour rolls are made - so could you 'reverse' this to apply to 'who got blocked' ?

Lots of responses - thanks! I will reply to them all in reverse order...

The 'Who's being blocked bit' - If they get twonked over then Armour rolls are made - so could you 'reverse' this to apply to 'who got blocked' ?

Indeedy I can... this is part of the as-yet-unwritten post on post-block rolls.

Very brief answer/suggestion as very busy, sorry.

The UKBBL uploader shows a table of “Best Dice in Block Totals”

So it shows, for example:
Player 1 (me) player 2 (daigaro ogami)
Skulls 1 42
POWS 19 0

I quite like that.

Hmm interesting. The number of block dice roll impacts this though doesn't it? ie more likely that your best dice is yellow if you roll 3 dice rather than 2. There is going to be a lot of data though so ideas like this on how to present it are welcome.

I think it's going to be very hard to determine a successful block due to the number of potentially desirable outcomes. The obvious example I can think of is surfing someone and therefore choosing to push them into a surfable position rather than knocking them down. You'll be able to track if a player gets surfed but what if it's a chain push to surf someone else?

I can't see any way of defining successful block beyond "didn't cause a turnover" and even that's situational (I take a 1d block which comes up skull against a skilless player to avoid a cloberring from a ClawPOMB monster, or a both down result to sack a ball carrier).

BBM already classifies them into success/failure/neither; I'm using that as the starting point. I agree that there are some niche examples that don't fit into these. However I think it's still worth pursuing and using a fixed definition of success/failure/neither. This is because the end goal of this (in my head anyway) is to see how lucky/unlucky you have been across the game. This will be done by comparing the total number of actual successes against the expected number of successes. An example....

Player with block gets 2D block on skilless player. Probability of success = 75%, or 0.75. Probability of failure = 1/36 (double skulls)
Skilless player gets 2D block on player with dodge, Probability of success = 11/36, or 0.30555. Probability of failure = 1/9

So here you have an expected block success count of 1.05555 and failure count of 0.138888, which is simply the sum of the probabilities. So you can compare your actual successes to these you can claim to be lucky, or unlucky. Now let's look at some of the examples you raised.

but what if it's a chain push to surf someone else?

I'd propose to ignore what you actually chose and use the "best" dice and use that to work out if it was a success or not. Doing this will also cover for when time runs out & the client picks one of your 2 block dice at random.

I take a 1d block which comes up skull against a skilless player to avoid a cloberring from a ClawPOMB monster

OK, so this would have a probability of success = 1/3. So if you "fail" it won't impact your overall totals much. And in any case, you can still claim that this one roll was lucky or unlucky, we are still capturing which one of those two (if any) has happened.

Remember, we are analysing the set of all blocks as a whole, so as long as we apply consistent methodology it doesn't really matter if not every line matches up with the coaches view of that block; the overall number will still be valid to determine if the player had lucky or unlucky block rolls. Also the number of unusual block rolls made in a game is probably going to be quite low so won't have a huge effect on the overall numbers anyway.

You may have noticed that I haven't covered rerolls here, which is also what daigaro referred to in the first reply on this thread. I will respond to this later this evening, or perhaps tomorrow... I will tackle 1 thing from that post though.

I.e. if you're blocking someone on the touchline with block and dodge then the 'positive' outcome chance is much higher than if they weren't on the sidelines; because defender stumbles and the two pushes now become more relevant

If you roll a yellow dice or pushed then the log will note a crowd-surfing has occurred. This is the most likely outcome... so *usually* we will get the correct probability. It's not perfect but it's better than BBM, which gives you block success/block attempt count with no context regarding if that was lucky or unlucky.

But did you give an option of what outcome was chosen? After a whoopin' last night by La I'd be interested also in some kind of 'overall' success Vs what you rolled - ie La rolled a double skull, so re-rolled into a skull and defender down. So, in the land of 'Chance' that was unlucky dice; but in the land of 'Nuffle' that was successful (and jammy). Does that make sense?

Alright, let me take you for a walk down Logic Lane...

At the first half of the street we have skill rerolls - things like Dodge, Sure Feet etc. When a coach rolls for one of these - and the player hasn't used the skill already, he gets two chances at it (we have to ignore Tackle here tbh unless the logs make clear that's what was being used), and the second chance is completely free of charge. So when Rav rolls for a GFI with his bull centaur who has Sure Feet, BBM says it was a 1 in 6 chance of failure. However we know, because of that skill, it's actually a 1 in 36 chance of failure (ie a double 1). So how to we model this? Well, it's exactly as you say. We have a roll success chance - which is usually straightforward to calculate, and then we have the concept of an "event" that takes things like this into consideration. One "event" can have multiple actual dice rolls, in this case the event was "a player tried to go for it" and it can consist of one or two actual rolls. Another place where I am planning on implementing this event concept is when trying to injure a player. It usually involves an armour roll and, if that passes, a casualty roll. However working out the chance of injuring/KO'ing is non-trivial, largely because of mighty blow, which can only impact on one of those two rolls. It's nice enough analysing the two things separately but combining them will lead to more reflective/better stats.

Now we get to the second half of Logic Lane, it's a bit darker and murkier, this is where team re-rolls live, as in your example. Now sure, in your case a re-roll was used and a success was had. However, there was a cost to this, a limited resource (a team re-roll) was used. One possibility is to treat every roll that does not have a skill re-roll as an "event" to model that there is an extra re-roll available (until it has been used for that turn, of course). However I don't think this is really fair as a carte blanche, when there are so many exceptions to the implicit rule here, that a team will use a re-roll if it fails any roll. For example, someone with Loner, my ogre team (which often has 6 really stupid rolls a turn, if one fails I usually shrug my shoulders and move on). If I failed the first 3 boneheads, and didn't re-roll them despite having a team re-roll, (1/6) * (1/6) * (1/6) is a lot closer to representing the bad luck I have had than if we assume I could've used the reroll at any time, which would give (1/36) * (1/36) * (1/36). Perhaps however, there is scope for applying this rule to any rolls which, if failed, would result in a turnover? This in turn has an assumption behind it that a player will do the least risky moves first and save their rerolls for the important things. Whilst this is sound Blood Bowl play, it's not what everyone does.

So basically, I've reached a firm conclusion on how to deal with skill re-rolls, but not team re-rolls. What do people think?

I would also be grateful if people could respond to the specific questions I raised in the opening post. Just look for the question marks.

So, in the land of 'Chance' that was unlucky dice; but in the land of 'Nuffle' that was successful (and jammy). Does that make sense?

BBM would show this as a neutral and a success for la. At the very least BBDB will call it a failure and a success. BBM only calls it a failure if a turnover resulted - which it didn't here. Also it wasn't that jammy (unless you had dodge), the chances of rolling at least one of the two yellow sides in 2 dice is 5/9.

I'd appreciate some more thoughts/responses to things I've raised here if possible?

Sorry, slowly catching up. I'm not sure if I have no more thoughts, or if I just have too much going on today (this is my 15 minute lunch...) to formulate any more thoughts. It may be that testing will reveal more concrete questions and responses?

(sorry if that sounds like a temporary cop out )