Why the optimizer sometimes stays put

Staying put is often the design working

You open the Optimizer screen, you check the coin your hashpower is mining, and you notice that it has not changed. Yesterday it was the same coin. The day before, the same again. The prices on the screen have moved, the per-coin numbers have shifted a little, and yet the optimizer has stayed put. It is natural to wonder whether something is broken. In most cases, it is not. Staying put is often the optimizer doing exactly what it was designed to do. This post explains why the optimizer sometimes stays put, how to tell a healthy quiet day from a real problem, and what to look for on the Optimizer screen so you can judge for yourself instead of guessing.

Reason 1: the 5% threshold

The first reason the optimizer stays put is the five percent threshold. ShiftHash evaluates twelve proof-of-work coins every fifteen minutes: BTC, BCH, LTC, DOGE, KAS, ETC, RVN, ZEC, DASH, ERG, FLUX and DGB. ETH is excluded because it is proof-of-stake, not mineable. Each evaluation scores every coin on estimated net yield in USD per watt per day, using live prices from CoinGecko and live difficulty from WhatToMine and mempool.space, minus a one percent pool fee. The optimizer then compares the best score with the score of the coin you are currently mining. It switches only when another coin beats the current pick by at least five percent of estimated net advantage. If the best alternative is two percent better, or three percent better, or four point nine percent better, the system stays put. That is not a failure to act. It is the threshold working as a noise filter.

Small differences are usually noise: a price moves a little, a difficulty reading ticks up or down, and the gap between two coins opens and closes within minutes. If the optimizer switched on every two percent gap, it would switch constantly, and most of those switches would be undone by the next evaluation. The five percent rule says a switch should only happen when the advantage is large enough to survive normal market wobble and clear enough that it is probably real. Below five percent, staying put is the better choice.

Reason 2: the daily switch budget

The second reason the optimizer stays put is the daily switch budget. Each plan tier has a maximum number of successful switches per day. The twenty dollar tier allows one switch per day, the one hundred dollar tier allows five, the three hundred dollar tier allows fifteen, the five hundred dollar tier allows thirty, and the one thousand dollar tier allows sixty. Only successful switches count, and the budget resets at midnight UTC. Limits stack across active contracts, so holding more than one active contract gives you a larger total budget for the day. The budget exists for the same reason as the threshold: to stop the system switching too often. Even a genuine advantage is worth less if the system spends its switches early in the day and has none left when a bigger advantage appears later. If the budget for the day is used up, the optimizer will wait until the midnight UTC reset, even if a new advantage above five percent appears. That can feel frustrating in the moment, but it is the system protecting its remaining switches for the next day rather than chasing every opportunity until it runs dry. The Optimizer screen shows switches used against the budget for today, so you can see whether the budget is the reason for the quiet. If the screen shows that you have used your full budget, the optimizer is not stuck. It is out of switches until midnight UTC.

Reason 3: a stale or missing feed

The third reason is the feed. The optimizer depends on live data, and live data sometimes fails. A price feed can go stale, a difficulty feed can go missing, or a reading can arrive too late to trust. When that happens, the system skips the evaluation rather than switching on bad data. Skipping is the safe choice, and it is an important part of the design. A switch made on a stale price could move your hashpower to a coin that is no longer the best choice, and a switch made on a missing difficulty reading could be based on a guess rather than a measurement. A missed evaluation costs far less than a wrong switch, so the system waits for fresh data and tries again at the next fifteen-minute evaluation. If the feed problem lasts for a while, the Optimizer screen will show it through the last evaluation time. A last evaluation time that is recent means the system is evaluating normally and simply finding no advantage above the threshold. A last evaluation time that is old, or a gap where evaluations should have happened, means the feed is stale or missing, and the system is correctly refusing to act on it. That distinction is the single most useful thing the last evaluation time tells you.

Reason 4: excluded coins

The fourth reason is excluded coins. Not every coin on the list is scored in every evaluation. FLUX is currently excluded because no reliable free difficulty and reward feed was found for it. DOGE is merged-mined with LTC, so its revenue folds into the LTC score and is never counted twice as a separate competing option. Excluded coins are listed on the Optimizer screen with the reason for their exclusion, so you can see what is left out and why. If you are watching FLUX and wondering why it never wins, the answer is on that screen: it is excluded until a trustworthy feed exists. That is not the optimizer ignoring a good coin. It is the optimizer refusing to score a coin it cannot measure reliably. The same logic applies to any other coin that is temporarily excluded for a feed reason. The exclusion is shown, the reason is shown, and the rest of the coins are still scored normally.

Reason 5: the lock

The fifth reason is the lock. You can lock the optimizer to a single coin, and locking pauses auto-switching entirely. While locked, your hashpower stays on the coin you chose, no matter what the evaluations say. The evaluations still produce information, but they do not move your hashpower. If you locked the optimizer some time ago and forgot about it, the screen will look as if the system has stopped working, when in fact it is doing what you asked it to do. Before concluding that anything is wrong, check whether a lock is active. If it is, and you want switching to resume, unlock it. If it is, and you want to stay on that coin, leave it locked and treat the quiet as expected.

Reading the Optimizer screen

Learning to read the Optimizer screen is the practical skill that ties all of this together. Start with the last evaluation time. It tells you whether the system is evaluating at all, and how fresh the data behind the current verdict is. A recent time means the verdict is current. An old time means the feed is stale or missing and evaluations are being skipped. Next, look at the best coin and its advantage against the five percent threshold. An advantage of half a percent means the coins are nearly tied. An advantage of four percent means a switch is near but not yet justified. An advantage above five percent, with switches remaining in the budget, should lead to a switch. Next, look at the per-coin scored yields, which show every coin's estimated net yield in USD per watt per day, so you can compare the coins yourself. Next, look at the excluded coins with their reasons, and the switches used against the budget for today. Finally, read the plain-language why-no-switch verdict. It explains in ordinary words why the system is staying put. If the box and the numbers agree, you have your answer.

A healthy quiet day

A healthy quiet day has a recognisable shape. The last evaluation time is recent, within the last fifteen minutes or so. The best advantage is below five percent, often well below it, and the per-coin table shows the coins clustered close together with no clear leader. Switches used is below the budget, because no switch was justified, not because the budget ran out. No lock is active. The why-no-switch box says, in plain language, that the advantage is below the threshold. Nothing is missing, nothing is stale, and nothing is locked. The optimizer evaluated, compared, and decided that staying put was the better choice. That is a healthy day. It may not feel exciting, but it is the system working correctly. Mining yield does not produce a large, clear advantage every day, and a system that switched every day regardless would be a system that had stopped filtering noise.

A day that should worry you

A day that should worry you looks different. The last evaluation time is old, hours old rather than minutes old, and it is not updating. The per-coin table is stale or missing. The best advantage is above five percent, switches remain in the budget, no lock is active, the feed looks fresh, and yet no switch happens across several evaluations. That combination is the one that deserves a closer look, because the stated rules say a switch should happen and it is not happening. A single missed switch is not proof of a problem, but a pattern across several evaluations, with fresh data and room in the budget, is worth reporting. The useful thing to note is what the screen showed: the last evaluation time, the best coin and its advantage, the switches used against the budget, whether a lock was active, and what the why-no-switch box said. Those five facts are enough to investigate properly.

Contrast: what a real switch looks like

Contrast helps here, because the optimizer does switch when the data justifies it, and the contrast with a quiet day is stark. On November 5, 2026, the engine made its first live auto-switch, moving allocation from BTC to ZEC at an eight hundred and forty-four point six zero percent estimated advantage. At that moment ZEC was scoring around zero point zero one eight one seven dollars per watt per day, while BTC was scoring around zero point zero zero one nine two dollars per watt per day. That is not a borderline case. It is not a four percent gap or a five point one percent gap. It is an advantage so large that the threshold, the budget and the feed all pointed the same way, and the system moved. That is what a switch looks like when it happens: a large, clear gap, fresh data, room in the budget, and a move. A quiet day is the opposite: a small gap, or no gap at all, and a decision to wait. Both are the same system working the same way. The switch and the wait are not two different behaviours. They are the same rule applied to different data.

Honest limits

There are honest limits to keep in mind while you watch the Optimizer screen. Every yield figure on it is an estimate, never guaranteed profit. Prices can move sharply between evaluations, difficulty can change as miners move between networks, and a coin that leads in one evaluation can lag in the next. The five percent threshold does not guarantee that a switch will be profitable, only that the estimated advantage at the moment of the switch was large enough to be worth acting on. A feed can go stale, and when it does the system will skip evaluations rather than guess. The switch budget, the threshold, the lock and the exclusions are all constraints that make the system more careful, not less capable. They trade a little speed for a lot of discipline, and that trade is deliberate.

Read the screen before worrying

So the next time the optimizer stays put, read the screen before worrying. Check the last evaluation time first. Then the best coin and its advantage against the five percent threshold. Then the per-coin table, the excluded coins, the switches used against the budget, and the why-no-switch box. In most cases, those six pieces of information will give you a complete answer in under a minute: the advantage is below the threshold, the budget is used up until midnight UTC, the feed is stale and evaluations are being skipped, a coin is excluded for a stated reason, or a lock is active. Those are all correct reasons to stay put. The optimizer is not a machine that must move to prove it is working. It is a machine that must decide, fifteen minutes at a time, whether moving is justified, and most of the time the honest answer is that it is not, and waiting is the right move.