A spell has a ten-second cooldown. You buy enough items to reach 50 ability haste. The next cooldown is about 6.67 seconds, not five.
Ability haste increases how quickly eligible abilities become available again. Each point adds 1% of the original casting rate, so 50 haste means a theoretical rate 50% above the starting rate. It does not mean you subtract 50% from the cooldown. That is why a bigger haste total can still leave you waiting when a short fight ends.
This guide covers modern PC Summoner’s Rift. The worked clocks below are calculations with explicit assumptions, not measurements from played games.
Turn ability haste into seconds
For an ordinary cooldown affected by the haste you are counting:
New cooldown = original cooldown × 100 ÷ (100 + haste)
Use the cooldown at the ability’s current rank before haste, not a cooldown already shortened by your build. For a ten-second ability with 50 haste, that is 10 × 100 ÷ 150 = 6.666... seconds.
The equivalent cooldown reduction is:
Cooldown reduction % = haste ÷ (100 + haste) × 100
Riot introduced this casting-rate approach when ability haste replaced cooldown reduction in patch 10.23. Those historical notes explain the stat; their old item builds and prices are not current recommendations.
| Haste | Equivalent cooldown reduction | A 10-second cooldown becomes |
|---|---|---|
| 0 | 0% | 10.00 seconds |
| 10 | 9.09% | 9.09 seconds |
| 20 | 16.67% | 8.33 seconds |
| 25 | 20% | 8.00 seconds |
| 50 | 33.33% | 6.67 seconds |
| 75 | 42.86% | 5.71 seconds |
| 100 | 50% | 5.00 seconds |
| 150 | 60% | 4.00 seconds |
Values are calculated, then rounded for display. The examples use unrounded cooldowns when counting casts.
Two independent sources of general haste add together. An item providing 20 and another providing 15 give you 35 haste before other modifiers, not two separate percentage reductions multiplied together. The current haste mechanics reference documents that additive rule and the distinction between the haste categories discussed below.
Check yourself: What haste would turn a 10-second cooldown into exactly 6 seconds?
The algebra gives 100 × (10 ÷ 6 - 1) = 66.666... haste. So the mathematical target is about 66.67, not 40. You wanted a 40% shorter cooldown, but haste and cooldown-reduction percentages are different numbers. A build offering whole-number haste would need at least 67 to reach six seconds or less under these assumptions.
Why another 20 haste seems to do less
From zero haste, adding 20 changes that ten-second cooldown to 8.33 seconds. You save about 1.67 seconds.
From 100 haste, adding another 20 changes it from five seconds to about 4.55. You save only about 0.45 seconds. The smaller saving is real. That alone does not tell you whether the purchase is worthwhile.
The theoretical casting rate is linear. Averaged over continuous repetitions, this spell has a rate of six casts per minute at zero haste. At 100 it is twelve; at 120 it is 13.2. That fractional number describes a sustained rate, not a spell you can partly cast. Each additional 20 haste adds the same 1.2 casts per minute to this model.
But the relative improvement is smaller when you already have more haste. Going from six to 7.2 is a 20% increase. Going from twelve to 13.2 is a 10% increase. The equal absolute gain is now a smaller share of what you already had.
For a purchase that adds haste to an existing amount:
Relative casting-rate increase % = added haste ÷ (100 + existing haste) × 100
This comparison holds the spell’s damage and casting conditions constant. If the new item replaces damage, mana, durability, or a useful passive, those changes need their own comparison.
The nine-second fight
Imagine a single ten-second ability is ready when combat begins. You cast at time zero, its cooldown starts immediately, and you recast as soon as it returns. The opportunity lasts until just before the nine-second mark. Ignore cast animations, travel time, mana, crowd control, cooldown refunds, and target access for this first calculation.
- With zero haste, you cast at 0. The next opportunity is at 10, after the window ends.
- With 50 haste, you cast at 0 and approximately 6.67.
- With 100 haste, you cast at 0 and 5. The next cast at 10 is still too late.
- With 150 haste, you cast at 0, 4, and 8.
Going from 50 to 100 haste has doubled the haste number, but it has not increased the cast count in this particular window. It has brought the second cast forward by about 1.67 seconds. That earlier availability could still be decisive if you need a shield or escape at five seconds.
Now change the encounter. The enemy retreats out of reach after your first spell and does not return until seven seconds. Both the 50-haste and 100-haste builds may take their second useful shot at seven. The lower cooldown existed, but there was no target in reach.
Check yourself: A fight ends before nine seconds. Does going from 50 to 100 haste guarantee a third cast of the ten-second spell?
No. With the first cast at zero and immediate reuse, both setups get two casts. The 100-haste setup gets its second one earlier. A third requires a different window, enough haste to shorten the interval further, or another mechanic such as a cooldown reset. Those possibilities are not included in this example.
Count the haste that belongs to the spell
Before adding the numbers, check which cooldown each one affects. General ability haste, basic-ability haste, ultimate haste, item haste, and summoner-spell haste do different jobs.
The item examples here use the ordinary Summoner’s Rift entries in Riot’s current item data, checked against their current client definitions. CommunityDragon hosts these Riot-origin client files. Arena and Classic items with similar names are excluded.
General ability haste: Fiendish Codex
Fiendish Codex currently provides 10 ability haste alongside 25 ability power. That general haste applies to eligible basic and ultimate cooldowns. It does not normally reduce summoner-spell or item cooldowns; those use their own haste categories unless an effect explicitly says otherwise.
Starting from zero other haste, its ten points turn a ten-second ability into about 9.09 seconds. That is useful to know, but it is not a reason to buy Codex on every champion. The ability power and the items it builds into still need to suit your champion.
Basic-only haste: Spear of Shojin
Spear of Shojin’s Dragonforce currently grants 25 basic ability haste. Its ordinary item entry does not list general ability haste in its base stats. Do not count those 25 points toward an ultimate.
If you already have 30 general haste, Shojin brings an eligible basic spell to 55 total haste. A ten-second cooldown becomes 10 × 100 ÷ 155, or about 6.45 seconds. Your ultimate still uses the original 30 general haste; Shojin adds none to that timer.
A fighter trying to repeat basic spells may value Shojin’s particular pool. Buying it primarily to bring back R sooner misses what Dragonforce offers. Its damage passive and the rest of your build still matter; this example isolates the cooldown benefit.
Ultimate haste: Malignance
Malignance currently grants 15 general ability haste plus 20 ultimate haste. With no other haste, an ordinary eligible basic spell uses 15, while the ultimate uses 35.
An otherwise unmodified 100-second ultimate would therefore have a cooldown of 100 × 100 ÷ 135, or about 74.07 seconds. Counting only the headline 15 would miss the ultimate-specific contribution. Applying the whole 35 to Q, W, and E would overstate what the item does for them.
The current Ultimate Hunter rune is another ultimate-only source: six haste initially, then five per Bounty Hunter stack earned from first takedowns on different enemy champions. Those stacks can bring the ultimate back sooner, but they do not speed up the basic spells you use between ultimates.
Summoner spells: Ionian Boots and Cosmic Insight
Ordinary Ionian Boots currently provide 10 general ability haste and 10 summoner-spell haste. Cosmic Insight provides 18 summoner-spell haste and 10 item haste, according to the current rune description.
With just those two summoner-haste contributions, the relevant total is 28. Flash’s standard cooldown is 300 seconds, so the isolated calculation is 300 × 100 ÷ 128 = 234.375 seconds, approximately 3 minutes 54.4 seconds.
The boots’ ten general haste is not another ten for Flash. Cosmic Insight’s ten item haste is not another ten either. Other bonuses, including role-quest rewards, and mode-specific rules are excluded. This is not a prediction of every player’s final Flash timer.
Item haste: the timer on an item effect
Cosmic Insight’s item haste is useful for a different set of timers. The mechanics reference distinguishes item cooldowns from champion-ability cooldowns and includes eligible active and passive item effects.
For a concrete example, the normal Zhonya’s Hourglass definition specifies a 120-second cooldown for its active. Ten item haste turns that into about 120 ÷ 1.1 = 109.09 seconds. If another Zhonya’s use is the goal, buying general ability haste does not solve this timer.
Zhonya’s 2.5-second stasis duration is not its 120-second reuse timer. Do not reduce both with the item-haste formula. An effect’s duration, a per-target restriction, or a special lockout may follow separate rules. Check the particular effect before applying the formula to it.
Cooldown ready is not the same as spell available
The calculations above assume the ability is governed by an ordinary haste-scaled cooldown. The cooldown formula does not describe every delay in a kit.
Do not apply the same percentage to a cast animation. And a finished cooldown still cannot supply missing mana, put a target in range, or end crowd control on your champion. An ability’s recast window or special cooldown rule also needs its own explanation. An ability with stored charges can have a recharge timer and a different restriction between casts. Do not label both numbers "the cooldown" and reduce both automatically.
The community-maintained League Wiki mechanics reference specifically excludes static cooldowns and target-immunity timers unless an effect says otherwise. Cooldown resets and reductions triggered by other actions are separate mechanics, not hidden amounts of ability haste.
This boundary also matters for tooltips. Read the relevant spell’s current rank and any special rules before substituting its number into the formula. A shop total alone cannot explain every timer in a transformed kit or a multi-stage ability.
A better shopping question than "How much haste is enough?"
Name the spell you are trying to get back sooner.
An ultimate-haste item can help prepare R for an objective without improving your basic-spell rotation by the same amount. A second basic shield during a dive needs a different haste pool. If the dive kills you before the shield returns, consider what would keep you alive for that second cast instead.
In a poke lane, a ready spell can sit unused because the mana bar is empty. Check that limit before paying for a shorter wait. For a brief burst exchange, compare the damage and safety of the spells you actually get to use. For an extended fight, repeated opportunities make the sustained-rate model more relevant, but only while you can keep casting.
A round haste total is not a build goal by itself. Start with the spell and its rank, add only the haste that applies to it, and compare the next purchase with the opportunity it can realistically create. In the nine-second example, the difference between 50 and 100 haste was not "two spells versus three." It was whether the second spell could happen at five seconds instead of 6.67.