All guides

Case studies: real success stories and failures

Posted February 26, 2026

Updated June 8, 2026

8 min read

Numbers, graphs, and lessons from real poker bot users. No names, but with specifics.

For whom: players considering bots; farmers who want to understand realistic expectations; investors who need numbers for their calculations.

Why case studies matter more than marketing promises

The poker bot market is full of claims about “guaranteed income” and “100% success.” Reality is more nuanced. The cases shown below are selected examples that went well; break-even and losing outcomes are common and are not shown here. Some users reach consistent profit within a month. Others blow their deposit in a week and blame the software, when the real problem runs deeper: a lack of understanding of the fundamentals, failure to follow recommendations, and greed.

This article is a collection of real stories. No embellishments. With numbers you can verify, and lessons worth learning before you invest your first dollar.

Success stories: when everything worked

Case #1: quick start on X-Poker

Parameter	Value
Platform	X-Poker, ClubGG, NLH
Fuel cost	~$200
Profit	~$2,650
ROI (on fuel)	~1225%

Context: A new partner started with a small fuel deposit on X-Poker and ClubGG (two of the softest club-based platforms). Within the first month of playing NLH at low-mid stakes, the bot generated ~$2,650 in profit. The partner ran multiple bot instances at the same tables, maximizing EV extraction from recreational players.

What the user did right:

Chose rooms with high fish traffic (X-Poker, ClubGG)
Seated multiple bots at the same table
Played at optimal stakes (didn’t chase high stakes right away)
Used TableSelect to choose tables
Had a sufficient stack

ROI above 500% is not the norm — it’s a lucky confluence of circumstances over a short sample. But $200 in fuel → $2,650 demonstrates the potential with the right approach. Note: ROI is calculated on fuel spend only; the initial one-time fee (from $700) is a separate investment that amortizes over time.

Case #2: scaling to a large farm

Parameter	Value
Platforms	ClubGG, PokerBros, X-Poker
Fuel cost	~$3,500
Profit	~$10,200
ROI (on fuel)	~+190%

Context: An experienced partner built a full-scale operation across ClubGG, PokerBros, and X-Poker. The farm ran dozens of bot accounts simultaneously, generating hundreds of thousands of hands per month. Account management was systematic: rotation every 7 days, residential proxies, proper bankroll allocation per account. The result: consistent 17 BB/100 winrate over 240K hands with ~$10,200 in profit on ~$3,500 fuel spend.

Sample	Profit (BB)	Comment
50K hands	+9,000 BB	Initial results, high variance
125K hands	+19,000 BB	Stabilization, EV converges with reality
240K hands	+39,000 BB	Long-term trend confirmed
Average win rate	17 BB/100	Above market average

ROI of +190% on fuel at scale is a more realistic benchmark for long-term operation. It’s not “breaking the bank in a week,” but a stable business with clear economics. Income depends on the number of bots, rooms, and stakes.

Case #3: the impact of stack depth on win rate

Context: Data aggregated from multiple partners across WePoker, HHPoker, and PokerBros. The AI’s performance was measured against stack depth to determine optimal bankroll per table. The results show a clear pattern — deeper stacks allow the AI to use its post-flop edge more effectively.

Win rate as a function of stack size:

Stack Depth	Win Rate (bb/100)
< 100 BB	~28
100-200 BB	~31
200-300 BB	~40
300-400 BB	~48
400+ BB	~47

The AI demonstrates maximum efficiency at stacks of 200bb+. The difference between short stack play (<100bb) and deep stack (300-400bb) is approximately 20 bb/100. Over the long run, that’s enormous money.

Recommended bankroll — starting from 50+ bb, optimally 100-200bb+ per account. If the stack grows to 400-500bb — it makes sense to lock in some profits and re-seat.

More about why the long run matters more than a single session result — in the article “Variance and the Long Run: Why Results Are Deceiving”

Data from real clubs: what the numbers show

Beyond individual user cases, we have aggregated data from our partners’ real clubs.

Daily results: ClubGG

Context: One partner’s daily results from a ClubGG club operation. The partner ran multiple NLH bot accounts at low-to-mid stakes. These are 9 individual playing days selected from a longer period. They show only profitable days. Losing days also occur and are a normal part of poker variance.

Day	Profit
1	+$2,383
2	+$2,548
3	+$3,039
4	+$3,978
5	+$4,012
6	+$4,194
7	+$4,620
8	+$6,607
9	+$9,103

Average profit across these 9 days: ~$4,500/day.

ClubGG club data showing daily player profit of +3,978 with 4,131 in total table fees

ClubGG club data showing daily player profit of +9,103 with 12,954 in total table fees

The “Fee” shown on these screenshots is the total rake generated across all players at the tables — not the amount paid by one player. Agents receive a percentage of this fee as rakeback, which is a separate income stream on top of player winnings.

Regional performance: ClubGG (Israel, ILS)

ClubGG clubs operate in multiple currencies worldwide. Here are results from an Israeli club: individual member statistics showing consistent profit over hundreds of hands:

ClubGG member detail showing +24,829 profit in 577 hands with Israel flag indicator

ClubGG member detail showing +19,529 profit in 308 hands in a single day

Large-scale operations: agent networks

Platform	Accounts	Hands/Week	Rake/Week
ClubGG (agent)	9	5,651	12,597 HKD
ClubGG (super-agent)	27	20,587	68,913 HKD
X-Poker	15–24	—	10,900–12,000 BRL
ClubGG / PokerBros	—	—	$16,900–26,300

All platforms use internal chip systems. The chip-to-currency rate is set by each club. ClubGG numbers above are from a Hong Kong-based operation (HKD). X-Poker uses Brazilian reais (BRL). PokerBros numbers are in USD-equivalent chips.

ClubGG agent statistics showing 9 downline accounts, 5,651 weekly hands, and 12,597 HKD in rake

ClubGG super-agent statistics showing 27 downline accounts, 20,587 weekly hands, and 68,913 HKD in rake

X-Poker agent panel showing total winnings of 23,595 BRL and 71,724 BRL in fees generated

PLO: session results across platforms

PokerBotAI bots support not only NLH but also PLO4, PLO5, PLO6, and OFC. Real PLO session results from our partners across multiple platforms:

+$23,072, +$75,121, +$135,224 — PLO5 100/50 on PokerBros
+$142,597, +$98,748 (PLO5 200/100–300/150)
+$62,432 in a single PLO6 50/100 session on ClubGG (with a 20,000 chip buy-in, 312% ROI per session)

ClubGG PLO6 50/100 table showing profit of +62,432 chips on a 20,000 chip buy-in

PokerBros session history showing three PLO5 100/50 sessions with profits of +23,072, +135,224, and +75,121

PLO5 results on Pokerrrr2 (India, INR):

Pokerrrr2 PLO5 200/100 session history showing three profitable sessions totaling +82,921 INR

High stakes: club games

At high stakes (blinds from 2,500/1,250 in club currency), results from a single session can reach hundreds of thousands and more. Top sessions from our partners: from +200,000 to +2,600,000 in local currency (e.g. Mongolian tugriks on Pokerrrr2).

Pokerrrr2 NLH 5,000/2,500 game results in Mongolian tugriks showing top player with +2.62M MNT profit

Pokerrrr2 NL Holdem 5,000/2,500 game results in Mongolian tugriks with top player earning +907,960 MNT

These impressive numbers are real, but with important caveats:

Currency. Large amounts on Pokerrrr2 and X-Poker are often expressed in local currency (tugriks, reais, etc.), not USD. The actual dollar equivalent can be several times smaller.
The field decides. Such results are possible where big money is actually circulating and there are enough recreational players. If your club has low traffic or small stakes, profit will be proportionally smaller.
Variance works both ways. Every winning streak can be followed by a downswing.

Losing periods: variance in action

Not all periods are profitable — and that’s a normal part of poker:

X-Poker, farm of 15-24 accounts: cumulative P&L was -36,392 BRL, while the farm generated 214,723 BRL in commissions. A loss on winnings with a gain on rake — a typical scenario on certain fields where rakeback compensates for variance.
PokerBros, weekly NLH series: out of 6 sessions — 4 profitable, 2 losing (-1,353 and -533). Weekly total: +8,975. Losing days are a normal part of the long run.
PokerBros, PLO4/PLO5: in a series of 12 sessions — 3 losing (-1,400, -900, -500), but the overall result was positive thanks to large winning sessions (+3,697, +2,086, and others).

Losing sessions and even losing weeks are not failure. They’re a normal part of poker. What matters is the result over the long run. If poker were always profitable — recreational players wouldn’t play.

PokerBros NL Holdem session history showing 6 sessions with 4 winning and 2 losing days

Weekly statistics from a PokerBros NLH player, 498 BB/100 winrate over 291 hands:

PokerBros player weekly statistics showing 498 BB/100 winrate and +2,898 profit in 291 hands

X-Poker agent panel showing 15-24 downline accounts with cumulative losses but 214,723 BRL in total fees generated

Major industry scandals: lessons for everyone

Martin Zamani: bot farm at the poker table (january 2026)

Well-known poker pro Martin Zamani was caught running a bot farm. This case showed that even professionals with a reputation can get detected. The consequences: loss of reputation, a ban, and a public scandal.

CoinPoker: $156k returned to victims

The CoinPoker platform conducted an investigation and returned $156K to players affected by bots. This is an example of how rooms respond to bot detection — fund confiscation and victim compensation.

PartyPoker: 291 accounts banned

PartyPoker blocked 291 accounts linked to bot operations. The mass ban shows that major rooms invest in detection systems and are prepared to take decisive action.

Lesson: improper setup and ignoring security measures lead to loss of accounts, deposits, and reputation. Following stealth recommendations is not optional. It’s a necessity.

Failure cases: where things went wrong

Failure #1: ignoring AI advice

Period	AI Adherence	Result
First week (~1,200 hands)	Manual play, overriding AI hints	Mediocre, negative EV
After switching to auto mode	100% (auto mode, multiple bots)	41 bb/100, consistent profit
Total	—	Time and money wasted in manual phase

The user played manually with AI hints for about a week (roughly 3-4 hours per day, ~1,200 hands total). They frequently overrode the AI recommendations, trusting their own reads over calculated lines. After a week of break-even results and growing frustration, they switched to full auto mode with multiple bot accounts. The difference was immediate. The bots hit 41 bb/100 with zero manual intervention.

At 60 hands per hour, manual play produces ~1,200 hands per week. The AI running in auto mode on multiple accounts generates that volume in hours — and executes every decision optimally, without fatigue or tilt.

Failure #2: datacenter IP instead of residential

A classic story: the user saved money on proxies, grabbed cheap server IPs. Within a week, they were banned.

Why this happens:

Poker rooms maintain databases of datacenter IPs
Multiple accounts from the same IP range = red flag
Even if the software works perfectly, a bad IP kills the account

Most bans happen not because of the software, but because of carelessness. Residential or mobile proxies are not optional — they’re a necessity.

Masking Best Practices + Launch Checklist

Failure #3: panic over short-term results

A real case from the statistics:

First 6,000 hands — zero result, the curve trends downward
The user panicked and started changing settings
Switched to another room, then came back
Ended up losing time and money from all the jumping around

Meanwhile, those who continued on the same track without changes:

By 20K hands — win rate of 38 bb/100
EV grew steadily, while actual results fluctuated
Variance smoothed out, profit materialized

6,000 hands is nothing. The minimum for evaluation is 50K+ hands. Judging a bot by two sessions is like evaluating the weather by a single day.

Failure #4: poor bankroll management

Most players who lose their bankroll do so not because of bad luck. They lose because they don’t manage risk properly.

A typical scenario:

Starting bankroll of $500 at NL50 (10 buy-ins)
Downswing of -3 buy-ins in one session
Attempt to recover at NL100
Result: -$500 in one evening

Standard approach — 40 buy-ins per stake level. For NL50, that’s $2,000 minimum. Drop to the next level down when you lose 10 buy-ins.

Lessons from the cases: what works

1. The long run decides everything

No success story is built on 5,000 hands. The minimum for conclusions is 50K+ hands. Over a short sample, variance can show anything: +50 bb/100 and -30 bb/100 alike. Only the long run reveals the true picture.

2. AI is smarter than intuition

The neural network is trained on 300+ million real hands (statistics from all poker rooms since the 2000s) and 7+ billion synthetic and solver-generated hands. Your intuition is based on a few thousand. When the AI says “fold” and you feel “call”, in 99% of cases, the AI is right. Failure #1 in this article proves it.

3. Infrastructure matters more than software

The best bot in the world won’t help if you’re using server proxies, holding 5 buy-ins at your level, and playing from one IP across 10 accounts. Most failures are not “bad bot” but “bad setup.”

4. Scale works

The large farm case shows: with the right approach, scaling linearly increases profit. ROI may decrease as volume grows, but absolute numbers increase.

5. Deep stack = more EV

The data is clear: the difference between playing at <100bb and 300-400bb is approximately 20 bb/100. The AI better capitalizes on its edge with deep stacks.

Realistic expectations: what you can expect

Scenario	Fuel Investment	Expected ROI (on fuel)	Monthly Profit
Single account	$200-500	200-400%	$400-2,000 (depends on stakes)
Small farm (5-15 acc)	$1,000-3,000	150-300%	$1,500-5,000 (depends on number of bots and scale)
Large farm (50+ acc)	$5,000+	100-250%	$5,000-15,000+ (depends on number of bots, rooms, and stakes)

These numbers are benchmarks, not guarantees. The one-time setup fee (from $700) is not included in the ROI calculation — it amortizes over time and becomes negligible at scale. Results depend on room selection, stakes, execution quality, and hundreds of other factors. Use the calculator at pokerbotai.net/est-profit for a personalized estimate.

More details in the article about bot ROI

Poker is a marathon, not a sprint. And bots play by the same rules.

Poker Bot ROI: Realistic Expectations

How Rooms Catch Bots: Detection Methods 2026

Variance and the Long Run: Why Results Are Deceiving