Decoding "noise" and BTC trends with neural networks (version with images)

This isn't the first time—and certainly won't be the last—that someone has attempted to predict Bitcoin's short-term movements based on historical data. It's a tough task, and perhaps we need to accept that the inherent chaos of this market makes precise prediction difficult, if not impossible.

For the past couple of years, I've been collecting data, processing it, and building various neural network architectures in search of a kind of philosopher's stone that could turn raw data into golden predictions. As you might expect, some of these networks were complete garbage—either failing to converge or suffering from horrible overfitting. However, I did manage to create several promising combinations that learned patterns, to an extent. While not good enough to track all market movements, these results suggested that some meaningful patterns might be present in the data I was feeding the networks.

The last two architectures I trained can generally follow market trends, but they are still not good enough. They particularly struggle during periods of high volatility. The two plots below show some examples, where the predictions were reasonable and the BTC price roughly oscillated around the trend lines indicated by the models (red lines). These are admittedly cherry-picked examples. The models' performance can deteriorate significantly during sharp market events.

BTC price evolution and prediction on December 17

BTC price evolution and prediction on December 22

As shown in the plots, while these models aren't reliable enough on their own, they might prove useful as supporting tools. I'm curious whether any of you have attempted similar projects and have suggestions or references to share. I'm not asking you to reveal your code or secret recipes 😉, just your thoughts and perhaps your frustrations if you've reached a point where you question whether more insights can be extracted from the data.

I will try to keep the model's outputs live in "BTC Live Forecast" (https://t.me/bitlivecast ). Feel free to have a look (there is no need to subscribe or pay anything).

Lightning Network to see considerable performance enhancement

Bitcoin's Lightning Network is set to see a considerable performance enhancement in the next week or two, with the release of the Fiber network from CKB. Solving scalability issues, Reducing transaction costs, Improving transaction speed, Providing multi-asset support, and increased interoperability.

Fiber/Lighting network illustration

INTRODUCTION TO FIBER NETWORK

By building off-chain channels on Nervos CKB, Fiber network aims to combine the successful experience of the Lightning Network with CKB's technical advantages to create a fast, low-cost, and decentralized multi-asset real-time payment network. Specifically:

1. Solving scalability issues: Through off-chain payment channels and multi-hop routing, Fiber Network can achieve high-throughput transaction processing, meeting the needs of large-scale users. 
2. Reducing transaction costs: By reducing the frequency of on-chain transactions, it lowers transaction fees, making micropayments feasible and efficient. 
3. Improving transaction speed: The instant confirmation of off-chain transactions provides a split second payment confirmation experience suitable for various instant payment scenarios. 
4. Multi-asset support: Fiber Network supports payments in a variety of digital assets, offering users a broader range of payment options. 
5. Interoperability: Fiber Network supports interoperability with the Bitcoin Lightning Network, providing support for cross-chain payments and asset transfers. 

👉 Announcing at the recent CKCon held in Thailand, that the Fiber network is in its second proof of concept phase and on its way to being released to Mainnet in the coming weeks.

👉 In previous offline events, the CKB team stated that they would set up 100,000 physical nodes in Fiber and the Lightning Network to promote the improvement and advancement of the P2P payment network.

INTRODUCTION TO RGB++

RGB++ is an extension protocol based on RGB, utilizing single-use seals and client-side validation technology to manage state changes and transaction verification. It maps Bitcoin UTXOs to Nervos CKB Cells via isomorphic bindings and uses script constraints on the CKB and Bitcoin chains to verify the correctness of state computation and the validity of ownership changes.

RGB++ addresses the technical challenges faced by the original RGB protocol in practical implementation and provides more possibilities, such as blockchain-enhanced client validation, transaction folding, shared states with ownerless contracts, and non-interactive transfers. It brings Turing-complete contract extensions and performance enhancements to Bitcoin without the need for cross-chain transfers or compromising security.

RESOURCES

Fiber Network website - https://www.ckbfiber.net/

Fiber Network White paper - https://github.com/nervosnetwork/fiber/blob/main/docs/light-paper.md

Github - https://github.com/nervosnetwork/fiber

Lead developer - https://x.com/crypcipher

RGB++ white paper - https://talk.nervos.org/t/rgb-protocol-light-paper-translation/7790

______________________________________________________________________________________

👉 TLDR: The Fiber network is set to be released in the coming weeks and aims to combine the successful experience of the Bitcoin Lightning Network with CKB's technical advantages to create a fast, low-cost, and decentralized multi-asset real-time payment network.

Stock Market Today: Nvidia Locks In $700M AI Play + Jeju Air Tragedy Sends Shockwaves Through Aviation

• Stocks hit the brakes on Monday as the market limped into the final stretch of 2024. The S&P 500 fell 1.1%, the Nasdaq dropped 1.2%, and the Dow slid 1%. A light news day left investors with little to rally around, and a lack of trading volume didn’t help matters either. Hopes for a last-minute Santa Claus rally? Fading fast.
• While stocks stumbled, Treasuries found some footing, with the 10-year yield easing to 4.54%. On the housing front, pending home sales rose for the fourth straight month, marking their highest level since early 2023. But with just a handful of trading hours left in the year, it seems the markets are wrapping up 2024 on a sour note.

STOCKS

Winners & Losers

What’s up 📈

• Pony AI jumped 11.15% after Guangdong allowed robocar tests across multiple cities with one permit, creating China’s largest autonomous driving zone. ($PONY)
• EQT gained 5.12%, driven by surging natural gas prices, which spiked over 17% Monday. Natural gas is on track for its best year since 2016. ($EQT)
• Nvidia barely made it in the green adding 0.35%, bucking the broader market downtrend and tech sell-off. The AI chipmaker’s stock has surged 185% year-to-date. ($NVDA)

What’s down 📉

• Archer Aviation dropped 10.02% after announcing plans to double its authorized Class A common stock from 700 million to 1.4 billion shares, as approved during a December 20 shareholder meeting. The company also greenlit the issuance of Class A shares to Stellantis under an August agreement. ($ACHR)
• Crypto stocks moved lower with bitcoin's decline: Mara Holdings fell 6.24%. ($MARA), MicroStrategy slid 8.19% after disclosing a sale of over 592,000 shares for $209 million, while also purchasing more than 2,100 bitcoins. ($MSTR), Coinbase declined 3.82%. ($COIN)
• Super Micro Computer lost 4.07% after terminating its financing and securities agreement with HSBC's Taiwan affiliate. ($SMCI)
• Technology stocks struggled, with the S&P 500 tech sector losing 1%: Tesla fell 3.30%. ($TSLA), Amazondropped 1.00%. ($AMZN), Apple declined 1.33%. ($AAPL), Microsoft shed 1.32%. ($MSFT)
• Boeing dropped 2.31%, as South Korea ordered inspections of all 737-800 planes following a deadly Jeju Air crash. ($BA)

Nvidia Locks In $700M AI Play

Nvidia has closed its $700 million acquisition of Israeli AI startup Run:ai, a move designed to optimize AI infrastructure and extend its dominance in the booming artificial intelligence sector. Regulatory hurdles? Cleared for takeoff—at least in Europe.

Opening the Playbook

Run:ai’s software, which helps maximize the efficiency of Nvidia’s GPUs, will now be open-sourced. Translation: competitors like AMD and Intel can use the software too. Why share the crown jewels? Nvidia’s move could be a play to quiet antitrust critics while expanding its influence across the broader AI ecosystem.

The startup, which Nvidia has worked with since 2020, adds a critical layer to Nvidia’s growing empire—helping power everything from data centers to cutting-edge AI systems.

Antitrust Drama

The deal wasn’t a breeze. Regulators in the EU gave the green light after a deep dive but didn’t miss the fact that Nvidia already controls about 80% of the GPU market. Meanwhile, the U.S. Department of Justice continues to take a closer look, adding a hint of uncertainty to Nvidia’s unchecked expansion.

For Nvidia, the stakes are high. As AI reshapes industries, its latest acquisition shows it’s not just chasing innovation—it’s determined to own the infrastructure behind it.

NEWS

Market Movements

• 🏦 Fannie Mae and Freddie Mac May Go Private: Billionaire investor Bill Ackman predicts that President-elect Donald Trump will remove Fannie Mae and Freddie Mac from conservatorship, potentially making them private companies again. Shares of Fannie Mae surged 18.4%, while Freddie Mac shares rose 18% following his announcement. ($FNMA, $FMCC)
• 💻 Major Telecom Hack Hits AT&T and Verizon: AT&T and Verizon confirmed they were targeted by the Chinese-linked Salt Typhoon cyberespionage operation, labeled the largest U.S. telecom hack, but assured their networks are now secure. ($T, $VZ)
• 🎥 Netflix Sets Streaming Record: Netflix set a U.S. streaming record with two Christmas Day NFL games,attracting nearly 65M viewers. The Ravens vs. Texans game peaked at 27M during Beyoncé's halftime show. ($NFLX)
• 🛒 Big Lots Store Deal Secures Brand Future: Big Lots secured a deal with Gordon Brothers to keep hundreds of stores operational after bankruptcy, with Variety Wholesalers planning to acquire up to 400 stores and two distribution centers. ($BIG)
• ☁️ IBM Acquisition Faces Scrutiny: IBM’s planned $6.4B acquisition of HashiCorp faces U.K. antitrust review, adding to scrutiny from the U.S. FTC amid concerns in the cloud sector. ($IBM, $HCP)
• 💼 Costco Defends DEI Initiatives: Costco pushed back against anti-DEI challenges, stating that diversity strengthens its workforce and customer appeal, urging shareholders to reject proposals opposing DEI initiatives. ($COST)
• 📈 Record ETF Inflows in 2024: U.S.-based ETFs saw record inflows of over $1T in 2024, boosting total assets to $10.6T (+30% YoY), driven by equity-focused funds, strong fixed-income demand, and bitcoin-linked funds. ($BTC)
• 🍔 2024 Consumer Trends Shape Retail and Dining: U.S. consumers prioritized value in 2024, shifting to discount retailers and fast food, reshaping retail and dining landscapes. Some businesses thrived while others struggled. ($WMT, $MCD)

Jeju Air Tragedy Sends Shockwaves Through Aviation

Boeing shares dropped 2% Monday after a tragic Jeju Air crash in South Korea killed 179 people, making it the deadliest aviation disaster in the nation’s history. Investigators are now looking into potential causes, including a bird strike and landing gear failure, as South Korea orders emergency inspections of all 737-800 planes in the country.

The Timeline

Here’s how the tragic event unfolded: Flight 2216, carrying 181 passengers and crew from Bangkok to Muan International Airport, declared a mayday after receiving a bird strike warning. The pilot aborted the first landing attempt, but during the second attempt, the landing gear failed to deploy. The jet skidded off the runway, struck a wall, and exploded into flames. Only two crew members survived.

Officials retrieved black boxes for analysis, while the U.S. National Transportation Safety Board and Boeing are lending support to the investigation. The 737-800 has been a reliable aviation staple, but this incident has raised serious questions.

Why It Matters

This isn’t just an isolated event—it’s a major moment for Boeing, which is still reeling from past controversies surrounding its 737 Max series. South Korea has 101 active 737-800s across carriers like Jeju Air, Korean Air, and T’way Air, and the crash has spurred a nationwide inspection to ensure compliance with safety protocols.

Adding to the scrutiny: Jeju Air’s stock hit an all-time low following the crash, and Boeing is under pressure to uphold the reputation of its long-standing workhorse model. Aviation analysts believe investigators will closely examine bird strike standards, emergency response protocols, and why the landing gear wasn’t deployed.

Bigger Picture: The tragedy also lands during political upheaval in South Korea, where Acting President Choi Sang-mok has declared a week of mourning. For Boeing and South Korean aviation, this is more than a PR crisis—it’s a test of safety practices in one of the industry’s most trusted aircraft models.

2024, my year in review: Lost a lot of $$, but learnt worth $$$$$$$

Story of a newbie options trader

(Warning: Long but interesting read ;) )

As the year draws to a close, I wanted to share my story, and my resolution for 2025. I hope you find the story interesting, and maybe we can learn something from each other.

I started learning and trading options in Apr 2024 ... the basics -- what are calls / puts? expiry dates, strike price ... learnt the jargon ... DTE??? ... LOL, I only knew DTF until then. Jokes aside.

I'm an engineer, and pretty good at math. I started loving options trading, because I realized that it is not really 'gambling', it is mathematics :)

Apr 2024: Trading
I was never an 'investor' mindset. I used to buy stocks low, and sell them when I get 10% profit. Then I tried options ... Started selling puts. This worked fine until the market started acting crazy from Aug. My puts got assigned, but fortunately had liquid cash. Waited for a while for the stock to go back up, but it didn't hence sold at a loss.
>>> First mistake: If you get assigned by shorting (cash secured) puts, use covered calls until you make profit, and then let the covered call get assigned, and exit the trade

Aug-Oct: Investor
Watched a lot of YouTube videos from prominent economists, CEOs from JP Morgan, Golman Sachs and the like ... Everyone unanimously said that the market was overvalued and was due for a crash. Buffet was sitting on a cashpile of $300 billion+ ... I bought puts hoping to make a shitload of money when the market crashes and then reinvest it back at the bottom ... and hopefully be a f*king millionaire by 2025. I really wanted to be an 'investor'
Surprise, Surprise!!! DIDN'T WORK!!!
>>> Second mistake: Never buy options (either calls or puts) without getting a CLEAR direction from the market. Without a direction, you're simply gambling! Never 'bet' only based on the views of experts -- they may be right -- but "markets can stay irrational longer than you can stay solvent". Learnt the hard way

After a lot of introspection, I started realizing that I can't really make money from options if I'm a "perma-bear" waiting for a black swan event to happen. Markets are designed to go up in the long run. No one wants them to go down. I started training my brain to become a bull. I came across an excellent quote -- "Bears sound clever, but bulls make money"

Nov: Bitcoin
My P&L had almost evened out by Nov, and I wanted to start afresh. I was hopeful of a Harris victory (because I lean left of center, but that's another matter). But when Trump started leading on Nov 5th, I decided I will be on the bulls side. I bought a lot of IBIT. I bought TESLA. It paid off. I am now trading options for IBIT and TESLA.

Then, I came across Microstrategy. I spent hours listening to various podcasters, critics, and folks on X. I realized that MSTR is a "trader's dream"! With its volatility, a good trader can really make a lot of money. And I did, I made thousands in profit buying MSTR call options!! Until ... I bought call options when it was at an ATH, and it crashed like a pack of cards. LOL
>>>> Third mistake: NEVER buy options when the volatility is high, and the stock is at an ATH or ATL. Wait for a direction.
Disclaimer: I am NOT promoting bitcoin or MSTR .. It could be applicable to any volatile stock, like TSLA.

Some other learnings:
-Unless you learn the greeks, you don't know what you're doing in options
-When the volatility is high, sell options
-When the volatility is low, buy options
-Wait for a trend
-When buying options, always prefer ones with the farthest DTE, and when selling options always consider the closest possible DTE to get the theta benefit.
-Preferably choose LEAPS (with deep ITM calls), and do covered calls
-Choose OTM strike prices only if you are REALLY sure. The risk is too high

In summary, I made a profit of around $105k in 8 months, and loss of $130k (My starting capital was around $140k). I consider it as an investment of $25k, and I think I learned a LOT in the last few months. I know I will continue to make profit. My resolution for 2025 is to minimize my losses. And some day I will be a f*kin millionaire ;)

Hope this helps someone. Wish you all a very happy new year!!!!