Do I understand correctly that the Stoic Classicist helps train up a student more cheaply than the usual lab option, 20 research per level, so 80 research over 4 actions to jump a student from new to expert?
Money Making on the wiki seems to suggest (at least for Cartographer’s Hoard) that the Stoic Classicist can be slightly helpful to EPA. I couldn’t get my head around how that would be so I ran some simulations on that grind. She is definitely not helpful if you use the otherwise optimal strategy. Assuming you retrain your Profound Student at 25cp (I cp short of hitting level 6 with its autofire -200 penalty), the best I could get in simulation without the Classicist was 99.531 actions per Cartographer’s Hoard (running 1 million hoard grinds) vs. with the Classicist 103.85 actions. If my understanding from above is correct, the issue is that you are basically spending 4 actions for no research gain, and that on average is bested by simply allowing the Profound Student to gain naturally with lower payouts (as you are always gaining at least something). Even if the Classicist simply bumped your student a level at no cost, it still somewhat surprisingly wasn’t cheaper to use her (102.22 actions). The only way she would be useful is if she gave some research each level bump.
Just wondering if I’m off somehow in my understanding of how she works, or if the "she is slightly helpful" sentiment is just intuition. Has anyone done formal numbers showing that she actually helps?
I thought that she trained 2 levels in 1 action. So 80 research over 2 actions, not 80 research over 4 actions. The main savings is probably the actions, not the research.
Thanks! Even assuming two level gains per action at a cost of 20 per use (so 40 research cost over 2 actions), she is still not useful for the Cartographer’s Hoard, coming in just slightly worse (99.664 actions) than the optimal non-Classicist strategy (assuming Hephaesta is in use). I suppose she might be useful on a longer grind like the Impossible Theorem where you have more student change-outs (there is on average just one retraining during the Hoard grind). I’ll have to check that at some point. For the moment I would conclude that the wiki claim under that section is not correct.
edited by Shaerys on 8/18/2020
Incidentally, the optimal strategy for Cartographer’s Hoard using Hephaesta is to maximize your watchful during training of the student only, but minimize it once your student is an expert. The idea is to keep your student in play without the need for retraining for as long as possible. The slight penalty for failures (2 research relative to a success) is easily offset by the reduced amount of actions wasted in training. I’d presume the same max/min strategy holds for Impossible Theorem and other longer grinds if the difference between success and failure is minor relative to total gain, but I haven’t formally checked those.
What chance of level up for Profound you are using in simulations? It very strange that its better to let them learn by themselves, you are losing a lot of research in opportunity cost, especially for first two levels.
I’ve been assuming it is a 50% chance of level up on any success, so two separate RNG checks, one for the pass/success, then a second check on 50% for whether a level up occurs on a success. Do we have better probabilities? I’ve been treating it as basically an alternate success rather than a rare success (as for the earlier students). Dunno if that is an accurate way to simulate or not…
You are only really losing out in opportunity costs the first couple of levels – at intermediate 3 and 4 you are still getting > 20 research. And levelling via the Classist has its own costs with a loss per action of 20. My watchful check is 85% at beginner (student level 1 and 2) and 83% at intermediate 3 and 4 (and minimizing at expert it is about 50%). The levelling up under those stats is evidently fast enough to outstrip the Classicist costs even given its better expert payoffs. Spending any action that doesn’t gain something just hurts. Even running the simulation with no costs on the level up – just the level gains per action – doesn’t perform as well under this simulation.
edited by Shaerys on 8/18/2020
Rerunning the sim with just a 30% chance of leveling up a student on a success does make the Classicist useful (assuming she jumps a student by two levels each time so that you need only utilize her option twice per Profound Student) with a revised average of 97.026 actions per Cartographer’s Hoard versus 98.603 actions without using her. So I suppose her utility comes down to what the actual probabilities are for student leveling.
edited by Shaerys on 8/19/2020