Asked DeepSeek poker strat. Here's what I got.
I asked DeepSeek, the open source AI program, what to do in a basic hand of poker. I wanted to see what it would say. I purposely left out key details, like how many seats at the table and what game it was and whether it was limit or NL.
My question:
playing a cash game of poker. effective stacks are 200 big blinds. I'm in the hijack with AT suited and the LJ raised first in to 3 big blinds. should I call, raise, or fold?
Here's the answer:
In poker, decisions like whether to call, raise, or fold depend on various factors, including your position, the strength of your hand, the tendencies of your opponents, and the dynamics of the table. Let’s break this situation down:
Key Details:
- **Your Hand**: A♥T♥ (Ace-Ten suited)
- **Your Position**: Hijack (HJ)
- **Opponent's Position**: LJ (Lojack) raised to 3 big blinds (BB)
- **Effective Stacks**: 200 BB
Considerations:
1. **Hand Strength**: A♥T♥ is a strong hand, especially suited. It has good playability post-flop due to its potential to make strong pairs, straights, flushes, and even nut flushes.
2. **Position**: You’re in the HJ, which is a relatively good position. The LJ raiser is in an earlier position, so you’ll have position on them post-flop, which is advantageous.
3. **Stack Sizes**: With 200 BB effective stacks, you have plenty of room to maneuver post-flop. This makes hands like ATs more valuable because of their implied odds.
4. **Opponent’s Tendencies**: If the LJ player is tight and only raises strong hands, you might want to proceed cautiously. If they’re looser, you can be more aggressive.
5. **Table Dynamics**: If the table is passive, you can consider raising to isolate the LJ player. If the table is aggressive, calling might be better to keep the pot controlled.
Options:
1. **Call**: This is a reasonable default play. By calling, you keep the pot small and see a flop in position. ATs plays well post-flop, and you can evaluate the situation based on the flop texture.
2. **Raise**: Raising to around 9-12 BB is also a viable option, especially if you think the LJ player is opening with a wide range. Raising can help you isolate the LJ player and take control of the hand. However, be prepared to play a larger pot, and consider your opponent’s tendencies before raising.
3. **Fold**: Folding is generally too tight here. ATs is a strong enough hand to play, especially in position, and folding would be overly conservative unless you have a specific read that the LJ player is only raising with premium hands.
Recommendation:
- **Call** is the most balanced and flexible option. It allows you to see a flop in position with a hand that has good potential to connect with the board. You can then decide how to proceed based on the flop texture and your opponent’s actions.
- **Raise** is also a good option if you’re confident in your post-flop play and want to apply pressure, especially against a weaker opponent.
- **Fold** is not recommended unless you have a very specific read that the LJ player is only raising with very strong hands.
In summary, **calling** is the most standard play here, but raising can also be a good option depending on your opponent and table dynamics.
*******
I'm going do a flop question next.
9 Replies
I realized I hadn't said what suit my hand is, so my next question was this:
i called. the flip is QhTh3c. I have AcTc.
the original raiser bets half pot.
what should I do next?
The answer:
Now that the flop has come **Q♥T♥3♣**, you have **A♥T♥**, which gives you **top pair with a strong kicker** and a **flush draw**. This is a very strong hand with a lot of potential to improve. Let’s analyze the situation and decide whether to **call**, **raise**, or **fold**.
---
Key Details:
- **Your Hand**: A♥T♥ (top pair, strong kicker, and a flush draw)
- **Flop**: Q♥T♥3♣
- **Opponent’s Action**: Bets half pot (likely around 4-5 BB into a pot of 8-10 BB)
- **Effective Stacks**: ~197 BB remaining
---
Considerations:
1. **Your Hand Strength**:
- You have **top pair with a strong kicker (Ace)**, which is a very good hand on this flop.
- You also have a **flush draw** (9 outs to a flush) and a **backdoor straight draw** (if the board runs out favorably).
- Your hand has a lot of **equity** against most of your opponent’s range.
2. **Opponent’s Range**:
- The LJ raiser’s pre-flop range likely includes strong hands like **AQ, KQ, QQ, TT, JJ, AK**, as well as some weaker hands like **KJ, QJ, or suited broadways**.
- On this flop, their half-pot bet could represent:
- **Strong hands**: QQ, TT, AQ, KQ (for value)
- **Draws**: KhJh, Jh9h, or other flush draws
- **Weaker hands**: A bluff or continuation bet with hands like AK, AJ, or KJ.
3. **Equity**:
- Against a strong value range (e.g., QQ, TT, AQ, KQ), you have ~30-40% equity.
- Against a wider range (including draws and weaker hands), your equity increases significantly.
4. **Stack Sizes**:
- With ~197 BB remaining, you have plenty of room to maneuver. You can afford to play aggressively or conservatively depending on your read.
5. **Position**:
- You’re in position, which gives you a significant advantage. You can control the size of the pot and make better decisions on later streets.
---
Options:
1. **Call**:
- Calling keeps the pot small and allows you to realize your equity. It’s a safe option if you’re unsure about your opponent’s strength or if you want to keep their bluffs in the hand.
- However, calling might allow your opponent to see a cheap turn card, which could improve their hand or kill your action.
2. **Raise**:
- Raising is a strong option here. You can build the pot with your strong hand and put pressure on your opponent. A raise could:
- Get value from weaker hands like KQ, QJ, or flush draws.
- Force folds from hands that have equity but aren’t strong enough to continue (e.g., AK, AJ).
- Protect your hand against draws (e.g., KhJh, Jh9h).
- A raise also sets you up to potentially win the pot on the turn if your opponent folds.
3. **Fold**:
- Folding is out of the question here. You have too much equity and a strong hand to consider folding.
---
Recommendation:
- **Raise** is the best option here. You have a strong hand with a lot of equity, and raising allows you to:
- Build the pot for value.
- Protect your hand against draws.
- Put pressure on your opponent’s weaker hands.
- A good raise size would be **2.5x to 3x their bet** (e.g., raise to ~12-15 BB). This is large enough to put pressure on your opponent but not so large that you’re committing yourself unnecessarily.
---
What to Do If They Call or Re-Raise:
- If your opponent **calls your raise**, you can continue betting on the turn if you improve (e.g., hit your flush or another Ten) or if the board becomes more favorable.
- If your opponent **re-raises**, you’ll need to evaluate their range and decide whether to call or fold based on their tendencies. However, with top pair and a flush draw, you’ll likely have enough equity to continue.
---
Summary:
- **Raise to ~12-15 BB**. This is the most aggressive and profitable line with your strong hand and equity. It allows you to build the pot, protect your hand, and put pressure on your opponent.
****
Notice I misspelled flop and it didn't matter. It did assume I had AhTh and not clubs, tho.
It also said I flopped tptk, but I didnt. Still, the advice it gave us pretty standard poker strat, even telling me my equity when I never asked about it or mentioned it.
Interesting.
Another question:
can you create a starting hand table for No limit Texas hold'em, 200 big blinds deep, eight handed. when open raising from the button.
Answer:
Creating a starting hand table for No-Limit Texas Hold'em (NLHE) is highly dependent on the specific strategy you want to employ, but I can provide a general guide for open-raising from the button with 200 big blinds deep in an 8-handed game. This table assumes a tight-aggressive (TAG) strategy, which is common for deep-stacked play.
---**Starting Hand Table for Open-Raising from the Button (200 BB Deep, 8-Handed)**
#### **Premium Hands (Always Open-Raise):**
- **Pairs**: AA, KK, QQ, JJ, TT, 99
- **Broadway Hands**: AKs, AQs, AJs, KQs, KJs, QJs, JTs
- **Suited Aces**: A5s-A2s (for their playability post-flop)
#### **Strong Hands (Open-Raise Most of the Time):**
- **Pairs**: 88, 77, 66
- **Broadway Hands**: ATs, KTs, QTs, J9s, T9s
- **Suited Connectors**: 98s, 87s, 76s, 65s, 54s
#### **Marginal Hands (Open-Raise Selectively):**
- **Pairs**: 55, 44, 33, 22
- **Broadway Hands**: AJo, KQo, KJo, QJo, JTo
- **Suited Gappers**: 97s, 86s, 75s, 64s, 53s
- **Suited One-Gappers**: T8s, 96s, 85s, 74s, 63s
#### **Speculative Hands (Open-Raise in Late Position Against Weak Opponents):**
- **Suited Aces**: A9s, A8s, A7s, A6s
- **Suited Kings**: K9s, K8s
- **Suited Queens**: Q9s, Q8s
- **Suited Connectors**: 43s, 32s
---**Key Considerations for 200 BB Deep Play:**
1. **Hand Selection**:
- Prioritize hands with good playability post-flop, such as suited connectors, suited aces, and pocket pairs.
- Avoid offsuit hands with low equity, like K7o or Q8o, as they are harder to play deep-stacked.
2. **Positional Awareness**:
- On the button, you have the best position post-flop, so you can widen your opening range compared to earlier positions.
3. **Stack Depth**:
- With 200 BBs, implied odds are higher, so speculative hands like small pocket pairs and suited connectors become more valuable.
4. **Opponent Tendencies**:
- Adjust your range based on the players in the blinds. Against tight players, you can open wider. Against aggressive players, tighten up.
5. **Post-Flop Play**:
- Deep stacks require careful pot control and hand-reading skills. Avoid overcommitting with marginal hands.
---Example Opening Range from the Button:
Here’s a visual representation of a typical opening range from the button (200 BB deep):
| **Hand Type** | **Example Hands** |
|------------------------|-----------------------------------------------------------------------------------|
| **Premium Hands** | AA, KK, QQ, JJ, TT, 99, AKs, AQs, AJs, KQs, KJs, QJs, JTs, A5s-A2s |
| **Strong Hands** | 88, 77, 66, ATs, KTs, QTs, J9s, T9s, 98s, 87s, 76s, 65s, 54s |
| **Marginal Hands** | 55, 44, 33, 22, AJo, KQo, KJo, QJo, JTo, 97s, 86s, 75s, 64s, 53s, T8s, 96s, 85s |
| **Speculative Hands** | A9s, A8s, A7s, A6s, K9s, K8s, Q9s, Q8s, 43s, 32s |
---
This is a general guide and should be adjusted based on your opponents' tendencies and your own skill level. Let me know if you'd like further clarification or help with specific scenarios!
I wonder if AI bots have access to online solvers, and this is just explaining solver outputs with words, or if they're just aggregating all of the online content available, synthesizing it all into something approximating a weighted average of all the pre-existing advice posted online.
If it's the former, we can just use a solver ourselves. If it's the latter, I suppose it's nice to have a bot find, read, and neatly summarize all the content out there, but I question if we're gaining anything we couldn't through our own study and logical reasoning, since that's what seems to drive poker theory, as delivered to us by theorists.
My experience using AI in other ways so far suggests it's a big time saver, but not necessarily a big game changer, in that it's not really generating new ideas or changing our understanding of existing ideas.
It can speed learning, but so far it doesn't add new knowledge.
I go back-and-forth on whether I should give a less trolly answer, but here's my short response:
To the extent that LLMs are useful for producing poker strategy*, you will have to do a LITTLE more work then sending a 200 character prompt and copy and pasting whatever it spits out. It takes some prompt engineering, proofreading, follow up prompts, etc.
*Which, at least for now, is limited to things like building out blog content or formatting a hand history.
I wonder if AI bots have access to online solvers, and this is just explaining solver outputs with words, or if they're just aggregating all of the online content available, synthesizing it all into something approximating a weighted average of all the pre-existing advice posted online.If it's the former, we can just use a solver ourselves. If it's the latter, I suppose it's ni
AI is an extremely broad term with a lot of specialized applications for smart people who are very thoughtful about it. PokerSnowie is built entirely on AI, and obviously that was quite revolutionary for poker when it came out. GTOW has AI-enabled functionality that (AFAIK) is extremely useful.
What AI is decidedly not useful for is typing lazy prompts into a mid-tier chatbot and copy and pasting the output to share with people who have specialized knowledge in the field as if it's going to provide us anything other than punchlines.
My experience with Chat GPT is akin to talking to an extremely polite and inhumanly fast personal assistant who has a passing knowledge of any topic, but makes frequent mistakes.
I also did a very in depth hand history with Chat GPT the other day and it made some very basic errors like confusing A high for top pair. I had to keep correcting it, but it was less useful in giving me insightful answers and more useful in helping me put my thoughts in order.
I wonder if AI bots have access to online solvers, and this is just explaining solver outputs with words, or if they're just aggregating all of the online content available, synthesizing it all into something approximating a weighted average of all the pre-existing advice posted online.
I kind of forgot to answer your question.
AI can do anything in this sentence, but what OP got was the latter.
DeepSeek is a specific type of AI, namely a large language model chatbot, which scrapes the publicly available web, aggregates it and gives a language-based response.
I am sure the AI functionality in GTOW is programmed to scrape and process the data of things like hand charts, as well as the frequency, EV, and equity views, etc and aggregate THAT, likely producing outputs that use GTOW's dashboards, views, reports, etc. This is extremely useful because while we can "just use a solver ourselves", most of what using a solver means is learning how to process a great deal of information and translate it into some framework that's digestible and implementable to our human brains (which is why different views and dashboards exist in the first place.)
To the extent that purely language-based interaction using publicly available information is useful, you do not have to just get a shotgun spray of randomly available information on the entirety of the web and can at least get it to analyze your HH as if it were UpSwing poker or submersible or what have you.
I kind of forgot to answer your question.AI can do anything in this sentence, but what OP got was the latter.DeepSeek is a specific type of AI, namely a large language model chatbot, which scrapes the publicly available web, aggregates it and gives a language-based response.I am sure the AI functionality in GTOW is programmed to scrape and process the data of things like hand c
It's all good, since my "question" was mostly rhetorical. Thanks for the courtesy of the attempt, though. You're alright, for a rando innerwebz stranger.
I think my point was that it seems like AI is at best a learning assistant, not a substitute for thinking. As you seem to have noted, we need to mind the inputs before we trust the outputs. If all we're getting is a summary of what AI finds on the web, it's may be like asking AI to summarize what's in a garbage dump (hint: garbage).
I'd think we'd benefit more from asking AI to explain the logical reasoning of solver outputs in words. That would at least aid our understanding of the logic a solver is using, rather than having various people using solvers attempt to explain it.
