Is there any reason you can't run the double and single lists MMA in parrelell?
When I say parrelell, i mean in series (double first then single) per pair comparison instance.
Since single list has some problems weighting heavily engineered bases with a single value, and the double is predicted to take significantly longer to match not balaced accounts, is it not possible to run the single and double lists in parrelell through the pool, or alternativly the entire pool through the double, then through the single inte its checker the entire pool using the double. For simplicity lets say the MMA will run both each time it selets both from the pool.
Two small adjustments will need to be made to each machmaker, one to speee up the search, the other to take away defenceless Max th11 and similar from having an advantage in the single list. Nerfs weight tax would be used for both adjustments.
to the double list matchmaker nerfs weight tax will be applied, but the value will be about a third as large as the weight tax used for single list matching. The double list would have 2 passes per calculation, if it fails on the first pass, it uses the figure plus weight penalty in the second pass.
So lets say a base had a weight of defence using the 2 list system of 15 000 and offence weight 120000. Lets say nerfs 2 list tax formula have us a defence weight if 60 000. The two list matchmaker would then first run its tests if the base weighed defence of 15000, if a match was found then the war was set, if a match was not found it runs again but this time with a defensive weight of 60000. If this time it finds it could match, it stores that knowledge but does not set the war.
If it doesnt match with the new number then of course the war is not set anyway. Iether way, if it could not match with the first pass of the double list, it then checks the clans using single list, for single list again nerfs weight penalty is applied to the base (i would prefer pro rata per clan possibly for single list but thats not important now), but his formulär will be different and will give a defence weighting of 110000 to that base that weighed 15000 defencesively.
The weight given will be such that the opposition will 98% of the time recieve iether parity or an advantage. The single list matchmaker could now run, and if it finds a a match then it sett the war, if not it also has a second pass with a defensive weight of 115000 for the defenclessish th11 (specific weights not important, illustrative). If the match is now possible, again it does not set the war, but stores the information (iether the values, or the clan identifier). If the match is not possible even with the weight tax, that is also noted. Then if the clans did not match on the first pass of the double and single list matchmakers, another clan is drawn from the pool is testet in the same way.
Then lets say no match is found after all clans have been testet on the first pass for iether matchmaker, the match will be set using the best match on the double list matchmaker second passess and if none exist the best match on the single list passes, if one exists. The war would then be set as normal as long as the other clan is still in the pool. And what if the second passes did not yield acceptable results, we have the option to return a no match was found after lets say 2 hours, or if chosen that process can be constantly repeated for each new clan that enters the pool, iether immedietly on a first pass or the best second pass checked every 10 minutes. Wether after some point one widens slowly the acceptable match criteria, or simply selects the best match available even if a mismatch at a cut off time, or goes on indefinetly, or returs a massage their was no match - which could possibly be determined by a war varible selection screen that clan leaders fill out in settings or for each war, as proposed by Dorsan amoungst others.
onyx and Dorsan, (And everyon else of course) what are your thoughts on using both the double and single list matchmakers to increase speed and closeness of match on average. Dorsan, the weight tax on the double list is not a weight correction, as that system doesnt need one, its a technique to speed up the time to match, with defenceless bases then simply paying an unfair penalty for the good of us all.
so its 4 calculation sets instead of 1 set for each pair comparison.
Tldr version: why not combine the single and double list matchmakers, run double list first, if it doesnt find a match, run single list. If no match found on single, increase the penalty weight for double and run again, if not match found, increase single weight tax again. If no match found after that iether widen parameters, or simply continue the search, or simply take the best of the worst, or return no match, as per each clans preference. Nerfs weight tax in the single list is to try weigh the base correctly, in the double list its try increase the likelyhood of ,and speed to, find a suitable match.
