Predra Efficiency Analysis

One thing I’ve noticed is very little discussion on what variant of the Predra Guerilla is better. Both are very similar except for stamina cost, SDR invade rate and MP per run.

The rates:

Predra 25 (P25) has a 1/20 or 5% invade rate for SDR.

Predra 50 (P50) has a 1/8 or 12.5% invade rate for SDR.

A simple look would suggest:

“Well if I can do 2 runs of P25 at 5% then I get 10% statistical chance per 50 stam for SDR compared to 12.5% and the latter is higher”

That’s how I thought of it at first until I went 7/87 on P50 and each run became gruelling because of the rates I’d been getting (even if it’s just some bad luck and I understand that). I decided to take another approach.

Below will be 2 analysis of farming P25 vs P50. They will always be in terms of 200 runs and 100 runs respectively (5000 stam total).

One of the assumptions is that SDR can be obtained at 12.5k MP which is the price on sale for skill delay resists.

Worst case scenario

Worst case scenario is where you just go 0/100 on P50 and 0/200 on P25. Furthermore, you go straight baby predras every run.

For both difficulties, the minimum is 300 mp (not counting coop mp because it’s equal for both difficulties). P25 is 4 floors, last being latent. The first 2 floors guarantee 100 mp predra OR SDR invade. Floor 3 is 100 mp, 500 mp or SDR. Similarly for P50, floors 3/5 are latents and floors 1,2 and 4 are predras OR SDR.

So in the worst case scenario we’re assuming 0 SDR invades and 100 mp drops.

In this case we’re looking at 200 runs x 300 mp/run = 60k mp for P25. This leaves you with 4 SDR at 12.5k each and 10k leftover.

For P50, you’re looking at 100 runs x 300 mp/run = 30k mp for P50. This leaves you with 2 SDR and 5k leftover.

Here the clear winner is P25 (though a very unrealistic situation).

Completely average scenario

You do 200 runs of P25 and get 10 SDR. Average. Nice!

A simple calculation of MP would be 100+100+(100+100+500)/3 or 433.33 mp per run on average. However this doesn’t account for the SDR spawns eating an mp floor.

To look at the loss of this situation, we already assumed that SDRs spawned 10 times and since they’re always on predra floors we can discount 1/3rd of the average mp per SDR invade or 144.44 mp lost on average per 1 SDR spawn and in this case you’d lose 1444 mp.

200 x 433.33 = 86,667 mp (rounded).

86,667 – 1444 = 85,223 mp in the end assuming perfectly average distribution of flr3 drops for predras and distribution of sdr spawn causing loss of mp.

This gets you enough to purchase 6 more SDR at 12.5k each with 10,223 mp leftover. Or .82 of the next purchase.

10 invades + 6.82 purchased = 16.82 sdr averaged

Now…

You do 100 runs of P50 and get 12.5 SDR. Average. Nice!

A simple calculation of MP would be 100+(100+100+500)/3 +(100+100+500)/3 or 566.66 mp per run on average. However this doesn’t account for the SDR spawns eating an mp floor.

To look at the loss of this situation, we already assumed that SDRs spawned 12.5 times and since they’re always on predra floors we can discount 1/3rd of the average mp per SDR invade or 188.88 mp lost on average per 1 SDR spawn and in this case you’d lose 2361 mp.

100 x 566.66 = 56,667 mp (rounded).

56,666 – 2361 = 54,305 mp in the end assuming perfectly average distribution of flr2 and 4 drops for predras and distribution of sdr spawn causing loss of mp.

This gets you enough to purchase 4 more SDR at 12.5k each with 4,305 mp leftover. That’s .3444 of an SDR.

12.5 invades + 4.34 purchased = 16.84 sdr averaged
P50 16.84 SDR per 5000 stam > P25 at 16.82 SDR per 5000 stam.

P50 is ever so marginally better when purely looking at SDR but you’re also looking at higher sensitivity to bad strings of runs.

It’s 85,223 mp on average for 5000 stam of P25 and 54,305 for 5000 stam of P50.

Other factors to consider:

1) Time. 200 runs of a dungeon roughly the same length as the one you’d be doing 100 runs of. Definitely consider this. It takes a lot less time to do the 100 runs of P50 compared to 200 runs of P25.

2) Investment. Because of the extra latent floor with 2 skill delay it requires an additional investment in either SDR on your buttons or inherits on buttons to counter delays. P25 only requires 4 SDR while P50 requires 6 SDR to startup (or pseudo SDR works).

3) Worst case scenario. We saw from the worst case scenario that you’re basically just twice as well off doing P25 if you’re going to get RNG trolled.

4) Above average rates. Getting above average rates on both difficulties by 1 extra per 100 runs sways the calculations a lot more heavily towards P25. The exact rates that Gungho has set is the point where both are basically equal all things considered (except for time and SDR investment as mentioned above) for SDR. However you’re sensitive to non-perfectly average outcomes and swings in either direction favor P25.

Conclusion

In the end given the sensitivity to RNG of P50, I’ve been choosing P25 over P50 when given the choice. However, time is very important to consider and P50 is by far more efficient in this respect.