Chip Up & Ship It (GGPoker, Spins recreational regular diary)

4K -> 20K Bankroll Challenge

Week 15

Start of Week 15 BR: 4,104 EUR
End of Week 15 BR: 3,000 EUR

Period Profit: -1,104 EUR
Tournaments Played: 155

Challenge Progress: -5% [IIIIIIIIII]

Summary:

- Back in the Game, but Running Cold: Finally got some poker action in with 155 tournaments played. Game quality was a mixed bag—not awful, but not exactly great either. Managed to end up on 2nd place in the weekly leaderboard, which softened the blow of some losses. That said, I’m running way below expected value, and the variance is stinging. Here’s a look at my weekly session:

- Grind Republic Kicking Off (Possible Rebrand Incoming): We’ve hit the ground running on Grind Republic, though we’re likely rebranding before launch to better capture the vibe. The SwongSim rework is coming along nicely—cracked the simulation logic and now focusing on meaningful visualizations. It’s fascinating to see how jackpots skew the data. Those 1M games are extreme outliers, happening once in a blue moon, but poker rooms hype them up in social media, lobbies, and news, creating a skewed perception that they’re more common than they are. Classic survivorship bias at play.

- Historical Leaderboards Analytics Module: We’ve started digging into the Historical Leaderboards Analytics module, with GGPoker as our first target since it’s the biggest room out there. Initial analysis and scraping of public APIs have turned up some intriguing insights:
- Over 6 years, GGPoker ran ~150K unique promotions, with most now concluded:
- The frequency and density of promotions have spiked recently, with daily promos becoming the norm:

and

- Hold’em dominates as the most promoted game type, but Short Deck surprisingly takes second place:

- Data Scraping Marathon: Ran a script to extract historical leaderboard standings last night. It was a beast—14 hours of runtime, downloading 28 GB of data across 152K files. Now comes the fun part: sifting through the noise to find what’s useful, transforming it, and loading it into a database for deeper analysis. This is where the data science magic starts to shine, and I’m pumped to uncover patterns that could give grinders an edge.

- Reflections & Next Steps: The poker grind was a reality check this week—variance is unfortunate, but it’s part of the game. Meanwhile, Grind Republic is picking up steam, and I’m juggling excitement and stress as we race toward the capstone deadline. The data we’re pulling is already sparking ideas, like how promo trends could influence play schedule or game selection. I’m still nervous about the MVP timeline and whether grinders will use it, but the poker mindset keeps me grounded: if it’s +EV, swing for it. Blogging’s back on track, and I’ll keep you posted on the project, the data science grind, and my eventual return to crushing the tables.

What do you think of the leaderboard analytics so far? Any ideas for features we should prioritize with this data? Tips for balancing poker variance with a crazy schedule? Drop your thoughts—I’m all ears.

Good luck on the tables

...

I tend to do a few hours training session on basic preflop ranges etc. on omaha cash if I have had week long break.

These are really great pieces of advice, folks - thanks! Indeed, I skipped my usual routines for the session: GTOWizard drills, reviewing past sessions, and reading Barry Carter’s Twitter for positive vibes. All this led to suboptimal performance.

Originally posted by Untiltable
It took me a while to understand the 0.2% number for MTTs, but I guess each day of a daily leaderboard counts as a promotion? And the comb-like spikes in the 5-year graph come from recurring weekly leaderboards?

Regarding the 0.2% for tournaments: promotions for MTTs are rare. The usual approach for MTTs is to offer big guarantees and occasionally add extra cash, but leaderboard-style promotions are uncommon. They do occur for some special series, which is likely where the 0.2% figure comes from. Out of 150,000 promotions, there were 300 MTT promos, mainly tied to the Bounty King Leaderboard and some WSOP promotions.

Regarding the number of promotions over time, I was misreading it - thanks for pointing that out. I’ve broken down the curve a bit. The "baseline" of the promotions curve consists of daily promotions.

Untiltable correctly noted the spiky parts - these are weekly promos, mostly from German Leaderboards. You can see a trial run in mid-2021 and active use in 2022. Since daily promos run on weekdays, the start of German weekly promos creates a spike, then it returns to the usual daily promo volume.

In late 2024, GGPoker added another set of weekly promos for PokerArabia, with starting days shifted by five days. This increased the density of the spikes:

Today, my head almost exploded due to the complexity and volume of data I tried to comprehend. By the end of the day, I decided to adopt a divide-and-conquer strategy, focusing solely on the German Spin and Go Leaderboards for analysis and limiting the time frame to start from January 1, 2023. Here are some interesting stats: Siggy79 is an absolute crusher of the German leaderboards, while your humble writer of this post sits in second place:

My next goal is to analyze the time series of the points required for each prize rank. I’ll try to identify any trends, seasonality, or predictor factors that influence the points needed to claim those crispy $ from the leaderboard and explain the variation:

We’re really excited to share that Poker Insights Lab is now live! Launching this site feels like a big moment for us. It’s been a journey to get here, and we’re happy to take this first step with you.

For now, we’re starting with the SwongSim remaster, our Spin&Go payout simulator, which is the heart of our first module. This tool uses the Monte Carlo method to simulate 10 different scenarios, running 1,000 games by default but comfortably handling up to 1 million games for deeper analysis. It’s a simple but powerful way to explore poker outcomes.

One feature we’re particularly proud of is something we haven’t seen elsewhere: a break-even cEV calculation based on payout structures and rakeback levels, computed using a binary search algorithm. It’s a handy tool for comparing simulations across platforms. For example, if you set the break-even point for $50 Spin&Go games on GGPoker and 888poker—both with similar rake levels—you’ll notice distinct run distributions due to their different payout structures, which creates noticeably different simulation outcomes despite identical EVs. It’s a small detail, but we think it adds real value for players who want to dig into the numbers.

Here’s a quick look at what we’ve built so far:

Our Starting Page: A straightforward entry point to explore the platform and get started with our tools.

Spin&Go Payout Simulator: The Spin&Go simulator in action, showing the interface for running simulations and analyzing results.

We’re in an active development phase for the next 3 weeks, focusing on delivering user value by developing and enhancing tools and capabilities. We’re very much interested in feedback and feature requests, so please share your thoughts to help us shape the platform.

If you’re curious, head over to https://pokerinsightslab.com/calculator and give Our Spin&Go Payout Simulator a try.

As always - have fun and luck on the tables!

...

Thanks for the kind words! I'm pleased to hear you loved the first release. Having worked on it for a while, I'm hyper-aware of things that could be improved and tend to focus on the imperfections. 😅 So, your positive feedback means a lot!

I'm eager to make it even better, so if you have any suggestions, please share them with me.

...

Thanks for sharing — adding tips to explain the fields is a great idea! And Primedope, I totally forgot about them; they’re good reference points for more tools. MTTs are trickier to model due to the ICM factor, but let me see how they approached it.

For the BRM, I was also thinking of incorporating it into the simulator. Besides profits/losses, we could model your deposit/withdrawal strategy. I followed an aggressive BRM because I was comfortable re-depositing if needed or moving down stakes. My bankroll on the platform was one part of a "global bankroll" that included other platforms and offline games. It was flexible, as I was comfortable borrowing from my overall budget. I’m still a big fan of aggressive BR management, but this should be revised if I move to higher limits.

Here’s what my calculator helped me realize: when you input top grinders’ monthly volumes (10,000 tournaments at a $200 BI level with weekly rakeback of 92% and 5 cEV), you should be comfortable with dips up to -$60,000 per week (that’s -300 BI). However, the long-run EV would be +$17,500 per month.

...

I have the following addition to your tool. It doesn't take into account the taxation of players from Germany, does it? Maybe you could add a checkbox for calculations for German players. On GGPoker, for example, every player has to pay an extra 3% to GGPoker.

Originally posted by Untiltable
ICM wouldn't come into play until stack sizes differ, so just looking at registration and likelihood of cashing should be a relatively straightforward prize pool distribution / field size calculation. The ROI adjustment is presumably similar (just with a lot more positions) to what you're doing in the calculator now. And then allowing for more prize pool options or even customization would be extremely useful - again, like you're already doing with the Spins.

The conversion from ROI to expected placements is not entirely clear to me. In Spins, it was more straightforward as I could calculate 1st place finish frequency based on cEV and average stack. The distribution of probabilities between 2nd and 3rd places was chosen somewhat intuitively. It's not clear to me how, in MTTs, the ROI performance indicator can be translated into placement probabilities, and this directly influences simulation logic. Nonetheless, I will think about it and consider how the implementation might look.

Originally posted by Kostja007
I have the following addition to your tool. It doesn't take into account the taxation of players from Germany, does it? Maybe you could add a checkbox for calculations for German players. On GGPoker, for example, every player has to pay an extra 3% to GGPoker.

This was in my initial scope but was dropped due to the added complexity and doubts about the share of the userbase that needs this feature. The complexity comes from the "special rules" of tax calculations. Rake is calculated very clearly:

((totalBuyIns - totalPayouts) / totalBuyIns) * 100

For the 3% German tax, you'd need to apply it to winnings only, so for 0 chipEV, it would be 0.93 * 0.03 = 0.0279. I could potentially include this tax in calculations, but then the user would need to enter a rakeback amount in the calculator, and rakeback should be based on rake with tax, or not? If with tax, how would the user keep track of the tax paid? For now, I’ve put this ticket back into the backlog and removed "GGPoker DE" from the payout structures list. I will reconsider if I receive more requests and data points for this.

An EVpool without Russian involvement would be great.
Unfortunately, the EVpools are firmly under Russian control.

...

Originally posted by PokerSpass
An EVpool without Russian involvement would be great.
Unfortunately, the EVpools are firmly under Russian control.

That's true, I share those concerns, but I'm not ready to launch my own pool yet It’s interesting to consider what it would take to start a pool. Probably gathering a few friends who can agree on the rules and trust each other to make transfers at the end of each period would be a good start.

Originally posted by Untiltable
I'm sure there are cleverer ways to do this as well. I don't really understand this stuff and couldn't do any of it without the AI tools, so I might be way off.

I'm confident we’ll be able to create something with AI assistance.

This week, we dove into a rabbit hole of leaderboard data analysis, focusing on Spin&Go daily leaderboards (GGPoker.com) across all stakes. We’re working to build a robust prediction model for the points required to enter each prize bucket on a given day. We did some feature engineering (identifying factors that influence the target value) and created a few graphs that I find enjoyable to explore:

We built our first models—XGBoost provided much better predictions (MAPE: 10-13% (Mean Absolute Percentage Error)) compared to a benchmark model using the previous day’s value (MAPE: 23%). We’re now exploring time series models (ARIMA) and neural networks, hoping to gain more insights that I’ll definitely share here.

Working on visualizing the player pool breakdowns. Discovered interesting heatmaps. The first shows Spin & Gold Leaderboard cash counts for the top-20 countries (heavily dominated by Russia). The second shows the Leaderboard prize sums for the top-20 countries (heavily dominated by Austria):

Played 3 evenings this week, took second place with a great points/prize money ratio.

Working with Sina on the Leaderboards Analytics module, set to launch by week's end. Looking forwards to share our progress!

Hit our second project milestone yesterday: 100 visitors on Poker Insights Lab. Many were friends and family unfamiliar with poker and Spin&Go simulators, but we saw surprising global traffic.

Stay tuned and good luck at the tables! :f_drink::f_drink::f_drink:

I’m happy to share the release of our new Leaderboard Analytics module app at https://leaderboards.pokerinsightslab.com !

Over the past two weeks, 2 of us has been working long hours to bring this module to life, and I’m excited to finally share the progress with you.

Current scope : GGPoker, Spin and Gold leaderboards only, data for the past ~5 months, both global and german leaderboards

Leaderboards Page: See the finished leaderboards rankings and dive into detailed stats.

Player Detail: Explore individual player performance with historical data.

Trends Analysis: see how requirements for min / max prizes in Leaderboards evolve with time

As always - we’d love to hear your feedback! Try it out and let us know what you think.

Happy grinding :f_drink::f_drink::f_drink:

...