A ‘swarm’ of beggars make a nice mob for instance. A ‘swarm’ of horses makes a nice stampede. These rather simple rules could be used to make ‘swarms’ of warrior level 1s into ‘squads.’

…dang… thanks for sparking this idea.

~runs off to flesh it out~

Assuming that the attack bonuses of our attack unit (A) are identical (as are any modifiers due to terrain, spells, equipment, etc), that the AC (plus modifiers) of the opposing unit (B) are also identical and that all members of unit A can make one attack per round. Assume that unit A has n members. The probability of single member of A hitting a member of B is p (which, thanks to D20 rules is always a minimum of 5% and a max of 95%) and the probability of a miss is 1-p.

The probability of unit A achieving r hits against unit B is:
nCr * p^r * (1-p)^(n-r)
where nCr | n! ÷ r!*(n-r)!

Example: unit of 5 people with a +3 modified attack bonus vs a unit with AC 12. Probability to hit is 55%.
Probability of unit achieving 3 hits:

5C3 * 0.55^3 * 0.45^2
=
((1×2×3×4×5)÷(1×2×3 × 1×2))*0.55*0.55*0.55*0.45*0.45
=
33.69%

From these figures you can put together a cumulative probability table to work out what you need to roll on a d100 in order to achieve any number of hits (dead easy to do in MS Excel or OpenOffice Calc. Apologies for the crappy ASCII table. I'll edit/repost it if necessary):

| n | r | nCr | p | 1-p | P(r hits)* | Cumu. Prob.** | Roll |
| 5 | 0 | 1 | 0.55| 0.45| 0.02 | 0.02 | 1-2 |
| 5 | 1 | 5 | 0.55| 0.45| 0.11 | 0.13 | 3-13 |
| 5 | 2 | 10 | 0.55| 0.45| 0.28 | 0.41 | 14-41 |
| 5 | 3 | 10 | 0.55| 0.45| 0.34 | 0.74 | 42-74 |
| 5 | 4 | 5 | 0.55| 0.45| 0.21 | 0.95 | 75-95 |
| 5 | 5 | 1 | 0.55| 0.45| 0.05 | 1.00 | 96-100 |

Excel formulae (for Calc exhange commas in POWER(a,b) with semi-colons):
row A | 5 | 0 | =FACT(A1)/(FACT(A2)*FACT(A1-A2)) | 0.55| =1-A4| =A3*POWER(A4,A2)*POWER(A5,A1-A2) | =A7 |
row B | =A1 | =A2+1 | =FACT(B1)/(FACT(B2)*FACT(B1-B2)) | =A4 | =A5 | =B3*POWER(B4,B2)*POWER(B5,B1-B2) | =A8+B7 |

…
copy and paste row b until the result in the second column matches that in the first.
*to 2 d.p so cumulative probability won’t always look right.

A table for the full range of value for p (you only actually need 5% – 50% in 5% increments (because at higher probabilities you can use the same table to work out the number of misses instead of hits) but you can go up to 95% if you like) can easily be put together. Here in the UK you’re actually given a booklet with these tables (amongst others) in to take into A-level Maths exams (YMMV – I’ve only sat papers from one exam board (AQA) so have no idea whether this is typical for others). From my copy (which was roughly 20 years old when I got it – we did get given new versions but had to hand those

back in) with probabilities converted to ranges (rounding to nearest whole number):

Unit of 5:
| no. of | prob of (hit) |
| hits | 5% | 10 % | 15% | 20% | 25% | 30% | 35% | 40% | 45% | 50% |
| 0 |1-77 |1-59 |1-44 |1-33 |1-24 |1-17 |1-12 |1-8 |1-5 |1-3 |
| 1 |78-98|60-92 |45-84|34-74|25-63|18-53|13-43|9-34 |6-26 |4-19 |
| 2 |99 |93-99 |85-97|75-94|84-90|54-84|44-77|35-68|27-59|20-50|
| 3 |100 |100 |98-99|95-99|91-98|85-97|78-95|69-91|60-87|51-81|
| 4 | | |100 |100 |99 |98-99|96-99|92-99|88-98|82-97|
| 5 | | | | |100 |100 |100 |100 | 99+ | 98+ |

Anyway, once you’ve got the number of hits, damage can be rolled normally (assuming same strength bonus to damage and same weapons) or you can come up with similar tables for each dice and add the appropriate extra damage on afterwards. Alternatively, you can assume average damage on a hit and round the total.

If you’ve got a unit larger than the table you’ve got handy, just roll on it multiple times so, for a unit of 30 you’d roll on the above table 6 times. Depending on the size of your units, you may find having tables for 5, 10 and 20 convenient or you can just multiply up the result of a roll on a smaller table. Obviously, at the extremes of probabilities and size of units, certain numbers of hits cannot be achieved using a d100. As a result, you may have to compromise on the size of table used or start rolling d1,000s, d10,000s or resort to a calculator/computer’s random number generator but you’ll have to create more accurate tables with the appropriate ranges.

Criticals can easily be taken into account. Say for a unit of 20 where you have five successful hits with a weapon with a threat range of 20, you can just roll a d100 and compare it against the 5% column in the table above, longsword would be the
10% column (assuming that a roll of 19 would hit otherwise use the 5%), rapier the 15% . Any threats can then be rolled individually as, for the most part, they’re going to be a 1 in 400 chance you shouldn’t get all that many or just use an appropriate size version of the table (again!) if you do randomly get a lot.

To keep things simple, assume that a unit retains it’s number of attacks regardless of it’s current hit points, ignore criticals and use average damage per hit. Best case, 100 attacks can be resolved with 5 rolls of a d100 whilst retaining the fundamentals of the d20 ruleset and a truely random number of hits.