Hold & Win slots get described in two ways that cannot both be true. One camp treats the respin round as a fixed mini-game: land six coins, get three respins, wait for a Collect, done. The other treats the whole thing as pure chaos, a coin flip dressed up in gold trim. Neither holds up once you look at how the feature is actually built. The respin phase draws from its own random number generator stream, separate from the base game, and the odds of landing a Collect symbol shift as more coins lock in place. That makes Hold & Win a system with states, not a one-shot gamble.
That distinction matters more than most players realize when they are choosing where to play. Imperial Wins accepts players from Canada and lists over a thousand titles, many of them Hold & Win variants from different studios, so the same feature name can hide very different math underneath. Two games can share identical visuals and paytable structure while resolving the respin phase at different speeds with different Collect frequencies. Knowing which lever a provider actually pulls is the difference between picking a slot on theme and picking it on how the feature behaves.
What follows is a mechanical walk-through rather than a hype piece. It covers how the trigger conditions work, why the separate RNG stream behind the respins changes what you should expect from a session, and how the Collect probability moves as the coin count climbs. From there, the model gets built out using active rows as the state variable, which gives a practical way to estimate how long a feature runs and what it returns on average. The last step applies all of it to real play: reading paytables, comparing volatility and return to player figures, and checking the licence line and cashier details at a Canadian-facing operator before you deposit a dollar.
How Hold & Win Slots Work: Core Mechanics and Feature Triggers
Hold & Win is a bonus format bolted onto a slot, not a slot type in itself. The base game spins like any other five-reel title, and the feature fires when six or more coin symbols land in a single spin. Those coins lock where they fall, the reels that held them go dark, and the remaining positions respin. Each new coin that appears resets the respin counter to three. The round ends when the counter hits zero or every cell on the grid is filled.
That is the whole skeleton. Everything else, and there is a lot of else, is in how the coins are priced and how the feature terminates. Most titles in the genre pay a fixed value per coin, usually a multiple of the total bet (1x, 2x, 5x, and so on). Some attach mini, minor, major and grand jackpot coins to the grid; those pay a flat amount rather than a bet multiple. A few, mostly from smaller studios, use progressive coin values that grow with each respin. The game library at a site like Imperial Wins passes a thousand titles, and the Hold & Win shelf within it spans all three variants, so the terminology matters before you compare anything.
Collect symbols, multipliers, and how the feature resolves
The Collect symbol is the mechanic that makes the feature a feature rather than a scatter payout. When it lands during the respin phase, it scoops the value of every locked coin into your balance and the coins clear. In most implementations the round continues with the respin counter reset, so a Collect is a payout event, not an ending. Multiplier symbols work differently: they land on the grid and apply a factor to the sum of locked coins at resolution, or they multiply the Collect payout itself. Read the paytable to see which. The difference is worth real money.
Three resolution paths exist across the genre:
- Grid full: all positions locked, often with a grand jackpot or a large fixed bonus attached.
- Counter to zero: the standard ending, with whatever coins are locked paid out.
- Collect with no respins left: some titles let a final Collect fire after the counter expires; others do not.
One myth worth killing here: the respin phase is not a continuation of the base-game spin. On audited titles it runs on a separate random number generator stream, drawn independently of the reels that triggered it. That matters for the next section, but the practical consequence is that base-game behaviour tells you nothing about what the respins will do.
Trigger frequency varies more than providers admit. A typical Hold & Win slot fires the feature roughly once in 250 to 400 base spins, with high-volatility builds pushing past 500. Return to player sits between 94% and 96.5% on most releases, and the feature usually accounts for 30% to 50% of that figure. If a paytable quotes a range, the lower number applies to the lowest bet level, a detail buried in the rules screen more often than it should be.
The opening package at Imperial Wins, CAD 1,000 plus 150 free spins, is priced in Canadian dollars, so coin values and bet multiples translate without conversion guesswork. Most slots also open in practice mode without an account, which is the cheapest way to watch a feature trigger and count the respins yourself before staking anything.
The Separate RNG Stream Behind the Respin Feature
Ask a room of players how the respin phase picks its symbols and most will say it is “the same reels, just frozen”. That is the first myth worth killing. In a properly built Hold & Win title, the respin phase is drawn from its own random number generator stream. The base game spins and the feature spins do not share a call to the same generator, and the outcomes are not seeded from the spin that triggered the feature.
Why split them? Because the feature is a different game. It has a different symbol set (no scatters, often no low pays), a different reel strip, and a different paytable. Forcing it through the base-game strips would either distort the base game’s return to player or make the feature unwinnable at sensible frequencies. Separate streams let the maths team tune each phase independently and keep the published RTP figure honest.
The practical consequence is that the trigger spin tells you nothing about the feature. A player who lands the sixth coin on a maximum-volatility spin and a player who lands it on a minimum-bet spin both enter the same probability space. There is no “hot” state carried over, no memory of how long the feature took to trigger, and no weighting that rewards a long dry spell. Anyone who tells you the feature “owes” a payout after a bad session is describing a machine that does not exist.
What the separation does change is the audit trail. Regulators and independent test labs sample the two streams separately. The lab runs millions of base-game spins to verify the published return to player, then runs millions of feature entries to verify the respin distribution, the Collect frequency, and the jackpot triggers. If the feature were drawn from the base stream, a single test would have to cover both and any drift in one would contaminate the other. Split streams make each phase testable on its own terms, and that is why you see licence lines from jurisdictions such as Malta, Gibraltar or Curaçao attached to studios that publish separate feature-level documentation.
The volatility story is where players feel the split most. Base-game volatility and feature volatility are not the same number. A slot can run a mild base game and a savage feature, or the reverse. The overall figure the provider publishes is a blend weighted by trigger frequency. In a typical Hold & Win title the feature might account for a third to a half of total return while occupying a small fraction of spins, which is exactly why the feature stream is tuned tighter and the payouts are lumpier. Two slots with an identical 96 per cent RTP can feel completely different because one puts the money in the base game and the other puts it in the respins.
One more myth: that the respin stream “knows” how many coins are already locked. It does not, in the sense of memory. Each respin is an independent draw from a distribution that the provider has already conditioned on the state of the board. The conditioning happens in the maths model, not in the generator. That distinction matters for anyone trying to model the feature, which is the subject of the next section.
For Canadian players the takeaway is simple. Treat the trigger as a door, not a forecast. Whether you are on a desktop or logged into a mobile session, the feature you enter is a fresh draw from a separate stream, and past results have no bearing on it. If you want to inspect the behaviour yourself, most slots open in practice mode without an account, so you can watch the respin phase run for free before you put a dollar of CAD at risk.
Why Collect Probability Shifts With Locked Coin Count
Most players assume the Collect symbol is drawn from a flat probability table for the whole feature. It is a tidy assumption and it is wrong. In the Hold & Win titles that publish their math models, the chance of a Collect landing on any given respin moves as coins lock. The direction of that movement is the part that surprises people.
Take a 5×3 grid with 15 cells. At the first respin, with one or two coins locked, the board is nearly empty. A Collect symbol that appears now has almost nothing to sweep, so designers weight it lower at that stage. As more cells fill, the same symbol becomes worth more, and the draw weights rise with it. The feature is not a static mini-game with a fixed hit rate. It is a system whose internal odds respond to its own state.
The mechanism behind this is straightforward once you see the code shape. The bonus round runs on a separate random number generator stream from the base game, and that stream is parameterised. Each respin, the engine reads the current locked-coin count, looks up a weight table, and draws from the symbol distribution for that row. Change the count and you change the distribution. Nothing about the draw is fixed except the table itself, which is audited and part of the game’s certification.
What does that look like in practice? Imperial wins casino is worth a closer look. A representative weight profile for a medium-volatility title might look like this:
| Locked coins (of 15) | Collect weight | New coin weight | What it means for the player |
|---|---|---|---|
| 1 to 3 | Low | High | Feature usually dies early; Collect is rare on purpose |
| 4 to 6 | Rising | High | Survival odds improve; most features end here |
| 7 to 9 | Moderate | Falling | Board is crowded; new coins get harder |
| 10 to 12 | High | Low | Collect is now likely; payout band widens |
| 13 to 15 | Very high | Very low | Grand jackpot territory; rare by construction |
Read the table as a trade-off, not a promise. The rising Collect weight at 10 or more coins is what makes the big wins possible, but the falling new-coin weight is what keeps them rare. Both curves are doing work. A title that lifts Collect odds without cutting new-coin odds would pay out too often and would fail its return to player target, so the two move in opposite directions by design.
This is where the active-row model earns its keep. Instead of tracking 15 individual cells, collapse the grid into the number of rows currently holding at least one coin. On a 3-row layout, that gives four states: zero rows active (the feature is over), one row, two rows, three rows. Transition probabilities between those states can be estimated from a few thousand logged features, and the resulting matrix tells you the expected number of respins and the expected return for each starting condition. It is a coarser model than a full cell-by-cell simulation, but it captures the dynamic behaviour that matters and it fits on a single page.
Two practical notes. First, the shift is not universal. Some providers keep a flat Collect weight and vary only the new-coin weight, which produces a different volatility profile and a flatter payout curve. Second, the shift is bounded by the licence line and the audit, so the weights cannot be tuned per player or per session. If a title feels like it is “due” a Collect after a long run of dead respins, that is memory bias in the player, not the math. The engine reads the board, not your history.
For Canadian players comparing titles at Imperial Wins, the takeaway is simple: the Collect symbol is not a coin flip you can bank on. It is a conditional draw, and the condition is how full the board has become.
Modelling the Feature: Rows, States, and Expected Value
To model a Hold & Win feature properly, you need one number that tells you where you are in the feature at any moment. That number is the count of active rows: the rows of the grid that contain at least one locked coin. Everything else follows from it.
The grid is a fixed 5×3 or 6×3 depending on the title, so the row count is a small integer with a handful of states. A row with no coin is inert. A row with one or more coins is live, meaning a Collect symbol landing anywhere in that row will sweep it. The transition from one state to the next depends on the respin outcome, which is drawn from the feature’s own stream, and the probability of adding a live row is simply the chance that a respin lands at least one coin in a previously empty row.
Building the Transition Table
Start with the per-cell coin probability from the paytable or the provider’s published rules, then compute the per-row chance of at least one coin: 1 minus (1 – p)^columns. Multiply across the empty rows to get the distribution of how many new rows go live on the next respin. From there you build a small transition matrix, state by state, and read off two things: the expected number of respins before the feature ends, and the distribution of the final coin count.
The Collect symbol adds a second layer. Because Collect probability rises with the number of live rows, the matrix is not symmetric. A state with four live rows has a much shorter expected remaining duration than a state with one, not only because there is more to collect but because the sweep is more likely to fire. That asymmetry is the whole point of the model. A static calculation that assumes a fixed Collect rate will overstate the value of early states and understate the value of late ones.
| Live rows | Typical Collect chance per respin | Expected remaining respins |
|---|---|---|
| 1 | Low | Long tail, often ends empty |
| 2 to 3 | Rising | Moderate, feature usually resolves |
| 4 to 5 | High | Short, sweep likely within a few spins |
Fill the cells with the actual numbers from the title you are studying. Providers rarely publish the Collect weighting directly, but you can estimate it from a few thousand logged features, or from the feature’s stated RTP contribution if the paytable breaks it out.
Expected value comes from folding the final coin distribution into the paytable: coin values, multiplier symbols, and any jackpot tiers. Multiply each terminal state by its payout and its probability, sum, and divide by the cost of the triggering spin. The result is the feature’s contribution to overall return to player, and it should sit within a point or two of the provider’s published figure if your transition estimates are sound. If it does not, your Collect weighting is wrong, not the provider’s maths.
One practical check: run the model on Imperial Wins game tiles where the feature rules are visible in the info panel, and compare the implied volatility to what you actually see in practice mode. Most slots open in practice mode without an account, so the test costs nothing but time. A model that predicts short features on volatile titles and long grinds on low-variance ones is doing its job. One that predicts the same duration everywhere is just a spreadsheet with opinions.
Evaluating Hold & Win Titles at Canadian Online Casinos
The model only pays off if you use it before you spin. On any Hold & Win title, open the paytable and the feature rules first and look for four numbers: the return to player, the base-game hit frequency, the minimum number of coins needed to trigger the respins, and the value of the smallest and largest jackpot symbols. A slot that quotes 96.2% RTP but needs six coins to trigger will feel very different from one at 95.5% that triggers on four, and the difference is not captured by the headline RTP alone.
Read the feature rules for the respin reset condition. Some titles reset the counter to three every time a new coin lands, which stretches the average feature length and pushes the session toward the top of the volatility band. Others run a fixed count of respins with no reset, which caps the feature at a short, predictable burst. The first type suits a bankroll that can absorb long dry spells; the second is easier to model because the number of draws is fixed from the moment the feature starts.
Compare providers on the same terms. The game library at Imperial Wins passes a thousand titles and splits into slots, table games, live casino and video poker, so Hold & Win releases sit alongside a lot of other content. Check the licence line in the footer and the named game providers before you deposit. A responsible operator publishes the RTP per title, not just a range, and list the playthrough clause for any bonus in plain language. Imperial Wins prices its cashier in Canadian dollars, which removes a conversion step from every deposit and withdrawal and makes the numbers on screen match the numbers on your statement.
The opening package of CAD 1,000 with 150 free spins is worth reading closely if you plan to run Hold & Win features on bonus funds. Bonus money usually carries a playthrough requirement, and high-volatility respin features can burn through a balance before the wagering is met. Decide in advance whether you want the bonus attached to a low-variance slot or whether you would rather play the feature with cash and keep the withdrawal clean.
The cashier lists more than twenty payment methods, including the Canadian payment methods most players already use. Withdrawals take roughly a day on an e-wallet and the better part of a working week on a bank transfer or card, so an e-wallet withdrawal keeps the loop short if you cash out after a good feature. Set a session limit before you start, and treat the practice mode as a free way to test a title’s trigger rate. Most slots open in practice mode without an account, which lets you watch fifty or a hundred features before risking a dollar.
Imperial Wins accepts players from Canada, so the account process is straightforward. The sign-up form requires one short form, and a safe Imperial Wins casino login means a unique password, two-factor authentication where offered, and an encrypted connection on your end. Never log in through a link from an email or a message board. Type the address yourself or use a bookmark. The Imperial Wins casino login page should sit behind HTTPS, and if a site asks for more identity documents than a standard verification, stop and check the licence line first.
Finally, keep a simple record. Note the trigger count, the number of respins you actually got, and whether a Collect symbol ever appeared on the last row. Twenty features is a small sample, but it will tell you whether the title behaves the way its paytable claims, and that is the whole point of building the model in the first place.
Before you stake a dollar on the next Hold & Win release, run the model on paper. Pull up the game in practice mode, most slots open without an account, and play the feature twenty or thirty times at minimum bet. Log how many coins lock on each respin, which symbols appear in the Collect and multiplier slots, and how often the round ends with nothing. That sample will not give you a precise return to player figure, but it will show you the shape of the distribution: whether the feature pays in small frequent hits or leans on rare jackpot symbols, and how long the respin loop typically runs before it resolves. Compare that against the published RTP and the volatility rating in the paytable, and you have a working estimate of what the slot costs per hour at your stake.
Then check the operator side, because the model is only as good as the cashier attached to it. Imperial Wins accepts players from Canada, prices its cashier in Canadian dollars, and lists more than twenty payment methods including the e-wallets and bank transfer options most Canadian players already use. Withdrawals take roughly a day on an e-wallet and the better part of a working week on a bank transfer or card, so factor that into your bankroll plan. Read the playthrough clause on the opening package of CAD 1,000 with 150 free spins before you claim it, confirm the licence line at the bottom of the page, and use the encrypted connection on the Imperial Wins casino login page rather than saving credentials on a shared device. The game library passes a thousand titles across slots, table games, live casino and video poker, and the game providers behind those tiles are named in the lobby, so you can match a studio’s reputation for Hold & Win mechanics to the numbers you just measured. A responsible operator will also give you deposit limits and a session timer in the account settings. Set them before your first spin, not after a bad session, and treat the feature as what it is: a separately seeded mini-game with shifting odds, not a guaranteed payout once the coins start locking.