Scaramouche has a passive called Gales of Reverie where the probability of activating wind arrows (the 'descent' effect) increases with each NA/CA hit. The base probability is 16% per hit. If a hit does not activate the wind arrows, the probability increases by 12%.
The question becomes: how does one calculate dps from this? What is the expected damage from this? It turns out this uses a markov chain. We will call this an example of a 'markov chain effect' (MC effect). The EO set from the chasm is another example of a markov chain effect.
To define the markov chain, start with the initial state 0. With probability .16 one can transition from 0 back to 0 (activate descent). With probability .84 one will transition to state 1 (with 12% increased probability to proc descent).
In state 1, with probability .28 one will transition back to state 0 (activate descent, reset probability). With probability .72 one will transition to state 2 (with 12% increased probability to proc descent). And so on.
From this, one can define a transition matrix P where \(P_{ij}\) is the probability of transitioning from state i to j. For the above this matrix looks like: $$P = \begin{pmatrix} .16 & .84 & \\ .28 & 0 & .72 & \\ .4 & 0 & 0 & .6 & \\ .52 & 0 & 0 & 0 & .48 & \\ .64 & 0 & 0 & 0 & 0 & .36 & \\ .76 & 0 & 0 & 0 & 0 & 0 & .24 & \\ .88 & 0 & 0 & 0 & 0 & 0 & 0 & .12 \\ 1 & 0 & 0 & 0 & 0 & 0 & 0 & 0 \end{pmatrix}$$
After 7 missed hits, the 8th hit is guaranteed to proc descent.
The above matrix P is an example of a stochastic matrix (where the rows add up to 1). What we want to find is the stationary distribution \(\pi \) of the matrix, solving the eigenvalue problem $$\pi = P^t \pi$$ Then \(\pi_i \) tells us the percent of time spent in state i. Then \(\pi_0 \) tells us the portion of transitions that activate descent. This involves solving the following system of equations $$(P^t - 1) \pi = 0$$ We also require the probabilities to sum to 1: $$\sum_i \pi_i = 1$$
The matrix equation gives the following system: $$\begin{align} .12\pi_6 - \pi_7 = 0 \\ .24\pi_5 - \pi_6 = 0 \\ .36\pi_4 - \pi_5 = 0 \\ \ldots \\ .84\pi_0 - \pi_1 = 0 \end{align} $$
Using the normalization condition: $$1 = \sum_i \pi_i = \pi_0 (1 + .84(1 + .72(1 + .6(1 + .48(1 + .36(1 + .24(1 + .12))))))) \approx 3.06 \pi_0$$ Solving this gives $$\pi_0 \approx .33$$ In other words, around 1/3 of the time we will activate descent.
Note that Wanderer's full NA combo involves 3 hits (technically the third NA is 2-hit, but we simplify for now). This means on average, at least one of those hits will activate the descent effect. The probability of not activating descent is around $$(2/3)^3 \approx .3$$ That is, there is around 70% chance to activate descent with a full NA combo.
Now at C6, each NA produces an additional hit, so the full NA combo involves 6 hits. This means on average the full NA combo can activate descent twice. The probability of not activating descent is around $$(2/3)^6 \approx .09$$ So there is around 91% chance to activate descent with a full NA combo.
The EO set from the chasm also uses a MCE. Via a similar method as that above, we have that roughly 1/2 of the time the effect is activated. In other words, the combat designers probably deliberately chose nice numbers.
Simulations can also be used to provide an estimate of the above effect. By repeating many trials, one can obtain estimates of things like how long on average it takes to activate the effect, the average dps, etc.