Pseudo-Random Distribution in Dota 2: How Crits, Bashes and Evasion Really Roll
Every time Phantom Assassin absorbs five hits in a row without a miss, someone in chat types “broken.” Every time Skull Basher lands three bashes back to back, the target calls the game rigged. Neither player is looking at the actual math — and that math is called Pseudo-Random Distribution, or PRD.
PRD is the probability system Valve uses for crits, bashes, and Maelstrom lightning in Dota 2. It deliberately changes the proc chance after each attack so that the stated percentage you read in the tooltip is accurate over time, but the pattern of procs is far less streaky than true random would produce. This guide covers every aspect of PRD in patch 7.41f: the formula, a complete proc math table for every major effect, which mechanics use it and which do not, and why the system still makes long streaks feel psychological even when the math says otherwise.
For the PA-specific interaction between Coup de Grace crits and Immaterial evasion, see our detailed breakdown at How to Counter Phantom Assassin: Evasion Math Explained. That page focuses on the counter-item strategy; this one covers the whole proc system.
In This Guide
What is Pseudo-Random Distribution and Why Did Valve Build It?
Pseudo-Random Distribution is a proc system where the probability of an event triggering changes after each attack based on whether that event has recently fired. In true random systems, every roll is independent: a 30% crit chance means each attack has exactly a 30% chance to crit regardless of history. PRD breaks this independence deliberately.
Valve implemented PRD to solve a specific problem with true random in a competitive game: variance. In true random, a 30% crit item can go 10 or 15 attacks without ever proccing, or can proc five times in six attacks. Both outcomes are statistically possible and both feel terrible to the players involved — one feels like their item is broken, the other feels like the enemy is cheating.
PRD does not change the average proc rate. A 30% crit item with PRD still crits 30% of attacks averaged across a long game. What PRD changes is the distribution. It compresses the tail ends of the probability distribution, making extreme droughts and extreme clusters much rarer than true random would produce. You still get streaks and droughts, but they are shorter and less severe.
The system was introduced in Dota 2 around 2013 and has been the standard proc mechanic for attack-modifier type effects ever since. It applies to crit items, bash items, and on-hit lightning effects. It does NOT apply to evasion, which remains true random — a distinction that matters significantly when analyzing a hero like Phantom Assassin.
The PRD Formula: How the Math Actually Works
PRD uses a value called the Coefficient (abbreviated C or PRD constant). After each attack that does NOT proc, the proc chance for the next attack increases by C. After a proc fires, the counter resets and the next attack starts at C again.
The proc chance on the Nth consecutive non-proc attack is: P(N) = min(1.0, C x N)
The coefficient C is calculated from the nominal proc chance p such that the expected proc rate across many attacks equals p. Solving for C given a desired average p requires summing the geometric series, which produces a non-linear relationship between C and p. A 30% nominal rate does not produce a C of 0.30 — it produces a C of approximately 0.0751 (7.51%). This is the starting chance, the chance on the very first attack after a proc fires.
What the Formula Means in Practice
Take Crystalys with its 30% nominal crit chance (C approximately 0.0751). On the attack immediately after a crit, the chance of critting is 7.51%. On the second attack after a crit, it is 15.02%. On the third, 22.53%. On the fourth, 30.04%. On the fifth, 37.55%. On the thirteenth consecutive non-crit, the chance exceeds 97%. By the fourteenth consecutive non-crit, it is guaranteed.
This means the maximum streak without a proc is bounded. For a 30% effect, you will never go more than 13-14 attacks without a crit, regardless of luck. In true random, you can go 20, 30, or theoretically unlimited attacks without a proc — just with decreasing probability. PRD caps the worst-case drought.
The tradeoff: PRD also caps the best-case cluster. You will not see three crits in a row as often as true random would produce. After each crit, you reset to 7.51% — which means consecutive crits require winning a low-probability roll twice in a row. The distribution shifts from “mostly average with extreme tails” to “tightly clustered around the average.”

Complete PRD Proc Math Table for Patch 7.41f
The table below shows every major PRD-governed effect in patch 7.41f, including the listed tooltip percentage, the starting PRD chance (C coefficient), and the maximum number of consecutive non-procs before the proc becomes guaranteed. PRD constant values are derived from the community-documented Dota 2 PRD formula (Dota 2 wiki, gamepedia PRD article).
| Effect | Listed (Nominal) % | Starting PRD Chance (C) | Max Non-Procs Before Guaranteed | Uses PRD |
|---|---|---|---|---|
| PA Coup de Grace (Deadly Focus proc) | 17% | ~3.6% | ~27 attacks | Yes |
| Skull Basher bash (melee) | 25% | ~6.0% | ~17 attacks | Yes |
| Skull Basher bash (ranged) | 10% | ~1.7% | ~59 attacks | Yes |
| Abyssal Blade bash (melee) | 25% | ~6.0% | ~17 attacks | Yes |
| Abyssal Blade bash (ranged) | 10% | ~1.7% | ~59 attacks | Yes |
| Crystalys crit | 30% | ~7.5% | ~14 attacks | Yes |
| Daedalus crit | 30% | ~7.5% | ~14 attacks | Yes |
| Monkey King Bar bash | 30% | ~7.5% | ~14 attacks | Yes |
| Maelstrom lightning | 30% | ~7.5% | ~14 attacks | Yes |
| Mjollnir lightning | 30% | ~7.5% | ~14 attacks | Yes |
| Wraith King Mortal Strike (lv 1) | 15% | ~2.9% | ~35 attacks | Yes |
| Wraith King Mortal Strike (lv 2) | 20% | ~4.5% | ~23 attacks | Yes |
| Wraith King Mortal Strike (lv 3) | 30% | ~7.5% | ~14 attacks | Yes |
| PA Immaterial evasion (lv 4) | 55% | 55% (flat, every attack) | No cap — true random | No — true random |
| Butterfly evasion | 35% | 35% (flat, every attack) | No cap — true random | No — true random |
| Bloodseeker Thirst | N/A — passive aura | N/A | N/A | No — not a proc |
| Battlefury cleave | 65% AoE damage | N/A — always triggers | N/A | No — always on |
Key reading of this table: effects listed as “Yes” for PRD have a low starting chance that grows with each non-proc attack, capping at 100%. Effects listed as “No” either roll flat true random each time (evasion) or always apply without any probability check (cleave, auras).

Effects That Use PRD in Patch 7.41f
PRD governs any per-attack effect that is described as a percentage chance to trigger an on-hit modifier. The consistent pattern: if the effect is “chance to deal bonus damage” or “chance to stun on hit,” it is almost certainly PRD.
Critical Strike Items
Crystalys — 30% chance to deal 175% crit damage. C approximately 0.0751. This is the first crit item most physical damage carries build, and the PRD coefficient means your first guaranteed-or-near-certain crit comes at the 13th or 14th consecutive non-crit attack. In a standard fight lasting 5-7 seconds against a high-attack-speed carry, you will see roughly 2-4 crits per engagement.
Daedalus — 30% nominal crit at 225% damage. Identical PRD coefficient to Crystalys despite the significant tooltip damage upgrade. Daedalus does not crit more often than Crystalys; it crits harder when it procs. Players who upgrade from Crystalys to Daedalus sometimes feel like crit frequency dropped — it did not. The higher burst from each proc just makes non-crit attacks feel more noticeable by comparison.
Bash Items
Skull Basher and Abyssal Blade — Melee users get 25% (C approximately 0.0602), ranged users get 10% (C approximately 0.0172). This explains why ranged heroes like Sniper or Drow Ranger feel almost no bash impact from Abyssal Blade — the 10% rate has a starting chance of 1.72%, meaning the guaranteed bash lands at roughly attack number 59. Most fights end before that.
For melee carries, the 25% melee bash rate is significantly better. The guaranteed bash arrives around attack 17, which falls within a typical teamfight duration at high attack speeds. Understanding attack speed breakpoints is directly relevant here: more attacks per second means you reach the guaranteed-bash threshold faster in actual fights.
Lightning Items
Maelstrom and Mjollnir — Both use 30% PRD for the chain lightning proc. The lightning is not an attack and cannot be missed, evaded, or blocked by most sources. It simply deals 150 (Maelstrom) or 200 (Mjollnir) magic damage to nearby targets. The PRD on the initial proc trigger means lightning arrives roughly every 3-4 attacks in extended farming or teamfight scenarios, which is why Maelstrom is such an efficient early farming item.
Monkey King Bar (MKB) — MKB provides True Strike on all attacks (bypassing all evasion checks) plus a 30% PRD chance for 160 bonus magic damage and a 0.01-second mini-stun per attack. The True Strike itself is not a proc — it applies on every single attack. The bonus damage is the PRD component.
Hero Ability Crits
PA Coup de Grace — In patch 7.41f, PA’s crit works through the Deadly Focus buff. She has a 17% PRD chance per attack to gain the Deadly Focus status. When Deadly Focus is consumed by an attack, that attack deals 200/325/450% crit damage at ability levels 1/2/3. The 17% rate produces a C of approximately 0.0358 (3.58% starting), meaning after a crit resets the counter, the chance to gain Deadly Focus starts low and climbs gradually. The maximum drought before a guaranteed proc arrives at approximately attack 27-28.
Wraith King Mortal Strike — WK’s crit passive uses PRD at all three levels. At level 3 (30%), the math is identical to Crystalys. Early levels (15% and 20%) produce longer expected gaps between crits, which is why early-game WK right-clicking in lane does not feel like a crit machine — the low-level rate has a high non-proc ceiling before the guarantee kicks in.
Effects That Do NOT Use PRD in Patch 7.41f
Several major mechanics are commonly confused with PRD but use entirely different systems. Knowing the difference matters for itemization and in-fight decision-making.
Evasion — True Random on Every Attack
Evasion in Dota 2 is NOT PRD. Every attack against an evading unit rolls its own independent probability check. A 55% evasion source means each incoming attack has a flat 55% chance to miss, with no memory of previous hits or misses. There is no accumulating counter, no guaranteed-miss ceiling, and no reset mechanism.
This is the source of much player frustration and confusion. Players who understand PRD sometimes assume evasion must also use it — it does not. A unit with 55% evasion CAN absorb 10 consecutive hits in a row (low probability, but possible). It CAN also miss 15 attacks in a row. Both are statistically legal outcomes that PRD would prevent for crit effects.
The practical implication: high evasion stacking (PA Immaterial at 55% plus Butterfly at 35%) produces an effective combined miss chance around 70.75% via the multiplicative formula, but every single attack is still an independent coin flip at that probability. There is no guaranteed miss window coming. For detailed evasion counter strategy against Phantom Assassin specifically, see our PA evasion counter guide — MKB and Silver Edge are the core answers.
Bloodseeker Thirst — Not a Proc at All
Bloodseeker’s Thirst passive increases his movement speed based on nearby enemy heroes’ current HP percentage. It is a continuously active aura effect that scales linearly with their health total — there is no chance roll of any kind. At full HP it provides minimal bonus; at very low HP it provides significant bonus. PRD is not involved because there is no binary “proc or no proc” event happening.
Battlefury Cleave — Always On
Battlefury’s cleave deals 65% of attack damage in an area around the primary target on every melee attack, automatically. The 65% figure describes the damage multiplier applied to the cleave area, not a proc chance. There is no roll. Every melee attack from a Battlefury carrier produces cleave damage. This is another effect that players sometimes misattribute to PRD because it “feels” like it has variance — it does not. All apparent cleave variance comes from target positioning and armor differences, not a proc system.
Spell-Based On-Hit Effects
Many hero abilities that trigger on attacks (like Bloodseeker Rupture, Viper Strike from Viper, or debuffs applied by spell-type attacks) do not use the PRD system. They use true random rolls or are guaranteed triggers. The PRD system applies specifically to the attack modifier category of effects — the physical on-hit procs from items and specific hero passives. Spell interactions follow their own probability rules, which are typically stated separately in the ability description.
Why PRD Streaks Feel Rigged: Math and Psychology
Even with PRD limiting drought and cluster lengths, players consistently report that crit and bash sequences feel unfair or rigged. There are three compounding reasons for this perception, and understanding them reduces tilt significantly.
Reason 1: High-Impact Events Create Stronger Memories
A 30% crit dealing 225% damage at 225 extra physical damage hits creates a visible, loud event: big number, animation, sound effect. Three consecutive non-crits that each deal 100 physical damage create no memorable event. Human memory is not a random-access log — it is a highlight reel. Players remember the cluster of two crits in five attacks. They do not remember the eleven non-crit attacks that preceded them and the twelve that followed. The perceived frequency of crits is far higher than the actual frequency because only the crits create memorable events.
This is a well-documented cognitive bias called availability heuristic: we judge the probability of events by how easily examples come to mind. Crits come to mind easily because they hurt. Non-crits do not come to mind at all.
Reason 2: Timing Amplifies Perceived Impact
PRD does not know when a fight starts. The counter continues across all attacks including farming. If you spend 40 attacks farming a jungle camp building up your PRD counter and then enter a fight, your first attack in that fight has a very high proc chance. From the defender’s perspective, you “instantly critted” at the start of the fight. From the PRD math perspective, you arrived with a full counter from 40 farm attacks. The timing amplification is real even if the math is fair.
Conversely, if you just critted on your last farming attack and immediately enter a fight, you start that fight with a reset counter (3.58% or 7.51% proc chance). The first five attacks in that fight might produce zero crits, which feels like a drought even though you just recently critted.
Reason 3: Multiple Sources Create Perceived Clustering
A carry with Crystalys, Abyssal Blade, and Maelstrom has three separate PRD counters running independently. Each counter resets independently on its respective proc. Maelstrom fires at attack 14 if it has not procced. The bash fires at attack 17. The crit fires at attack 14 again. These ceilings can line up, and when they do, you see lightning, bash, and crit in a very short window — which feels like a impossible cluster even though each one is mathematically expected from its own counter.
The independence of these counters is also why stacking multiple crit items (Crystalys + Daedalus on the same hero) doubles the damage per crit but does not double crit frequency. Each item tracks its own counter. Both can fire on the same attack (proc checks are independent), but both can also go extended non-proc runs simultaneously.
How PRD Affects Your Build Decisions
PRD math has direct implications for itemization, particularly around crit stacking, bash timing, and the transition from farming to fighting.
Crit Stacking: Diminishing Frequency Returns
A single Crystalys gives 30% crit. Adding a Daedalus gives a second 30% crit source. But because each rolls independently via PRD, the expected crit frequency across 10 attacks does not double — it increases modestly because either item can independently proc on any given attack. The damage per crit increases dramatically (Daedalus crits hit harder than Crystalys), but the frequency gain is much smaller than intuition suggests.
This is why high-level PA, WK, and carry builds typically do not stack two Daedalus items. The first Daedalus dramatically increases crit damage. The second Daedalus adds only marginal crit frequency because both items compete for the same attack slot and their combined expected proc rate is far below 60% per attack. Other item slots provide better damage returns than a second crit item with the same PRD coefficient.
Bash Timing: Commit After the Reset
Because bash PRD counters carry over between targets, there is a real strategic advantage to initiating on a target when your bash counter is built up. The first attack on a target has whatever accumulated PRD count you had on your previous target. Players who swap targets frequently during fights are effectively throwing away accumulated bash counters.
For carries with Abyssal Blade who want to commit bash procs, focusing one target for multiple attacks builds the counter efficiently. Switching targets resets nothing — PRD tracks your hero’s counter across all attacks — but staying on one target means every attack contributes to that target’s guaranteed bash window. Understanding high-MMR fight positioning starts with this kind of mechanical depth.
Maelstrom as a Farming Tool vs Teamfight Tool
Maelstrom’s 30% PRD lightning proc is its most valuable feature during farming because camp clearing generates many attacks in quick succession, producing 2-3 lightning procs per camp at higher attack speeds. In teamfights, the same proc rate applies, but teamfights have fewer total attacks — you might get 8-12 attacks in a 5-second fight, producing approximately 2-3 procs. Maelstrom’s farming efficiency is real; its teamfight contribution is real but proportionally similar. The item’s core value is the attack speed it provides alongside the lightning, not the lightning alone.
Attack Speed and PRD: Reaching the Guarantee Faster
Higher attack speed does not change the PRD coefficient. The starting chance stays at 7.51% and climbs by 7.51% per non-proc attack, regardless of how fast you attack. What attack speed changes is how quickly you reach the guaranteed-proc threshold in real time. A hero attacking at 1 attack per second reaches the guaranteed Crystalys crit at attack 14 after 14 seconds. A hero at 3 attacks per second reaches it in under 5 seconds. In a 5-second teamfight, the slow attacker might not see a guaranteed crit window at all; the fast attacker will. This is one of the underappreciated reasons high-attack-speed carries feel more reliably threatening in fights — they spend more time in the “high PRD counter” zone where procs are increasingly likely, rather than resetting frequently. See our expert coaching breakdowns for how this translates to specific hero build paths.
Applying PRD Knowledge by Bracket
Lower brackets rarely think about proc systems at all — they react to crits and bashes as random events and treat them as luck. Higher brackets internalize the PRD math intuitively even without naming it. The jump from reacting to understanding is the entire gap in how these brackets approach physical damage heroes.
Herald to Crusader: Recognize the Pattern
The most useful single takeaway at this bracket is that evasion is true random and crits are not. When you build MKB against PA, you are not “getting lucky” — you are guaranteeing hit rate on every attack while she rolls individual dice for each miss check. You are also knowing that after her last crit, her next crit probability resets to 3.58%. The window immediately after a PA crit is the safest fighting window to stay in. This understanding alone changes how you play teamfights against her.
Archon to Ancient: Exploit the Counter Timing
At these brackets, players start tracking fight initiation relative to item procs. A Abyssal Blade melee carry that has not procced bash in 10+ attacks is about to proc. Initiating a fight now, rather than waiting, gets you the guaranteed bash early in the engagement. This is not obvious from looking at the item — it requires knowing the PRD counter is near its ceiling. The same logic applies to timing your own fights when you know your bash or crit is due.
Divine to Immortal: Multiple Counter Awareness
At the highest brackets, players track approximately how deep their PRD counters are on each item simultaneously. After a fight where Daedalus critted once and Abyssal bashed twice, the Daedalus counter is closer to ceiling than the bash counter. The next fight starts with an early crit window but a longer bash window. Fight length targets, target selection, and item timing are all influenced by this counter state. MMR climbing at these brackets often comes down to exactly this level of mechanical precision.
Turn Mechanics Knowledge Into MMR
Understanding PRD is the theoretical side. Applying it under pressure, in real ranked games, against opponents who are also optimizing — that is where coaching bridges the gap. TeamSmurf’s pro staff coach players from every bracket, with session focus areas as specific as itemization timing around proc mechanics.