Samrat Batth, PhD, ATSR

Source: Notion | Last edited: 2023-10-30 | ID: edbd6cb4-6ee...

---

Video

• The meeting is for understanding the client’s needs and offering technical assistance.
• Integration and technical setup support are mentioned as possible services.

Samrat’s Setup

• The client’s model has long and short positions based on predictions.
• Exposures are calculated and delta is determined every five minutes.
• The delta is traded in the market for the next few minutes.
• Market making orders are generally used for passive fills.

Integration Request

• The client asks if they can send their deltas for execution to the assistant’s service.

Final Clarification

• The client also asks if the assistant’s service supports pegging orders to mid prices.

Trading Execution

• Internalization of buy and sell orders can avoid the need to send orders to the market.
• Sending deltas to the trading platform to execute trades can benefit both parties.
• The execution algorithm can handle the trade execution
• Slippage is currently at 0.53 basis points, which is considered very good.
• Maker and taker fees are included in the slippage number.
• The slippage is measured from the arrival price when the signal is received.
• The ratio of maker to taker trades is 1:1.
• The slippage varies for different coins, with smaller coins having higher slippage.

Data Access

• Historical market data is available, including tick level data and bars.
• Real-time pricing feeds can be accessed.
• There is a wide range of symbols available for key line data, but only BTC is available for tick by tick data.
• The order book data includes 1000 bids and 1000 asks per tick.

Trading Contracts

• Around 25 contracts are traded simultaneously.
• Each contract has its own order size and execution.

Trade Execution API

• API access is required to send trades to the trading platform.
• Order size, reference price, and target time frame can be provided.
• The execution order can be optimized to target the arrival price.
• A basket order is sent every 5 minutes with signals for 25 different contracts.

Trade History

• An API is needed to retrieve the trades executed by the trading platform.

Real-Time Execution Feedback

• Real-time feedback is necessary to calculate the correct delta in a streaming execution system.
• Feedback is important to ensure accurate exposure calculation.

Exposure vs. Delta

• Exposure can be a better metric to use than delta in calculating trades.
• By providing the full exposure rather than just the delta, the correct trades can be executed.

Access to Real-Time Data

• Real-time data can be accessed via a WebSocket connection.
• Data is fetched from Binance and Coinbase and stored in a database.

Accessing Historical Bar Data

• Access to historical bar data for analysis and prediction is required.
• The current focus is on writing a WebSocket connection to improve speed.

Access to Execution API

• Access to the trading API is needed to send signals for execution.
• Signals can be sent from any server location.

Server Infrastructure

• The server is co-located in AWS Tokyo.
• The database is located in DynamoDB in the US West 2 region.

Signal Transmission Preference

• Prefer to send signals via WebSocket for ease of use and reliability.
• Database sync can be prone to issues, whereas WebSockets are simpler.

Slippage and Volume Considerations

• Turnover is around 20 times a day with a holding period of approximately one hour.
• Market volume is considered in prediction calculations to avoid high slippage.

Preferred Data Vendor

• IQ feed is one option, but it may not have historical data.
• TRTH is another vendor, but its pricing is uncertain.

C2C Contract

• C2C is a corp to corp contract.
• LLC to LLC is the standard approach.

Performance-based Infrastructure

• Start with a base of 30% performance fee.
• Increase fee by 5% for every 200K made.
• Fee increment is tiered for different earnings ranges.

Evaluation of Net Asset Value (NAV) Increase

• Net asset value increase can be evaluated by comparing buy and sell signals.
• Orders can be combined and re-marketed to reduce fees.

Priority for Trading API

• Obtaining a trading API is the highest priority.
• Data-related tasks can be addressed later.

Trading API

• Priority is to have a trading API for sending exposures
• Data optimization can be delayed

Performance Fee

• Excel table will be shared to discuss performance fee
• Awaiting feedback on the fee arrangement

Timeline and Handshake

• Discussed final handshake timing
• Expecting a final outline next week
• Further discussion required on timeline

Video

Infrastructure Discussion Delayed

• The technical team is having a delay in discussing infrastructure.
• They need one or two more days to wrap up their work.

Product Sharing Sheet Feedback

• The main conflict is the 50% performance allocation.
• 30% is the maximum received from outside investors.
• There might be arrangements that offer a higher payout.
• The justification for 50% and how to proceed with institutional investors are important points to discuss.

Capacity Constrained Strategy

• The strategy is capacity constrained, limiting the target P&L.
• Investing externally would require a higher target P&L to make sense.
• The goal is to run externally for six months and then evaluate becoming an internal employee.
• Take home percentage with investors has been slightly higher than 50%.
• Multiplier and absolute P&L are important factors to consider.

Terry Suggested: Lower Capacity and Performance Allocation

• Lowering the capacity allows for adjusting the performance allocation.
• Lowering the amount invested won’t impact the strategy’s performance significantly.
• Exploring the possibility of running with a lower capacity to accommodate outside money.

Samrat Supposed: The total capacity for this strategy is 3 million.

• If 3 million is invested with a 50% return, it will result in a 1.5 million profit.
• Reducing the risk to 2 million will yield a lower absolute profit of 1 million.

Samrat Rebutted: Running the strategy with lower capacity hinders scalability.

• Running the strategy with multiple investors on different accounts can lead to adverse selection and higher slippage.
• Executing trades from different accounts adversely impacts execution prices.

The current setup is exclusive and not ideal for running with multiple investors.

• The capacity constraint signal limits the ability to run the strategy with multiple investors.
• Adverse selection and higher slippage occur when trading from different accounts on the same exchange.

Adding additional parameters to the performance allocation.

• Setting the parameter to true introduces additional parameters in columns AA and AB.
• The allocation depends on the investment time horizon, rewarding faster gains and longer holding periods.

Adjusting the performance allocation based on the investment time horizon.

• A tiered structure will be implemented to fine-tune the allocation based on the return timeframe.
• The extreme case of 180 days with 50% return may not be justifiable anymore.

Considering the option of a part-time role and compensation.

• Formulating the part-time role into the agreement for consultants.
• Further discussion needed on compensation such as hourly rate.

Exploring internalization to improve execution and reduce costs.

• Internalizing the flow can eliminate the need to execute trades in the market and pay spreads.
• Executing trades internally can improve fill rates and reduce adverse selection.

Adverse selection and higher costs when trading the same direction on the same exchange.

• When trading the same direction on the same exchange, adverse selection occurs, leading to higher costs.
• Internalization can mitigate adverse selection and improve trading performance.

How to do it effectively

• It is important to have clear communication and understanding
• Setting up a convenient time is crucial 2023-07-18

Video

Adjustments in the Spreadsheet

• The starting and result rate have been adjusted in the spreadsheet.
• Additional adjustments have been made by Terry Li.

UPA and Investment Cycle Adjustments

• The UPA in SNRF Arrear Cycle has been adjusted to six from five.
• The CI% has been adjusted from 50% to 42%

Recruiter Payment and Investment Time Horizon

• Samrat has some questions regarding recruiter payment but it’s not Samrat’s concern since it is not coming out of the Performance Allotment (PA)
• The agreement mentions a maximum acceptable investment time horizon (MAITH) of 28 days, but it’s not a concern because the recruitment agent Michael of Cerebro Search is not following a fixed schedule but the Head Hunter Retention (HHR) that’s based on new ITHE reached and new Settlement triggered.

Review of the Agreement

• Samrat has gone over the agreement and finds it generally okay.
• The agreement will be reviewed by the Samrat’s lawyer for any language issues.

LLC Setup and Infrastructure Fee

• An LLC will be set up for consulting purposes.
• There will be a monthly infrastructure fee of $5,000, but the exact cost will depend on the agreed infrastructure setup.

Discussion on Infrastructure Cost

• Samrat needs more information about the infrastructure setup to provide a proper estimate of the monthly cost.
• A meeting with the tech team will be arranged to discuss this.

Hybrid Role Engagement

• Samrat is open to a part-time engagement to help build the macro infrastructure.
• A commitment of 20 hours a week and an hourly cost will be discussed.

Next Steps

• Samrat suggests finalizing the agreement and Google Spreadsheets once the infrastructure cost and part-time engagement details are agreed upon.
• Samrat is available to go online and finalize the infrastructure part.
• Terry suggests that further discussions are needed internally among Eon’s team members before a green light is given.

---

Video

Document Review

• Samrat is reviewing the document and will revert with questions via email.

• Specific language points will be highlighted for discussion。 Excel Sheet

• The Excel sheet is discussed, specifically the column labeled UPA.

• UPA stands for ARIA, which is the Safety Net Reserve Fund. SNRF Explanation

• The SNRF is a first loss scenario that absorbs drawdown beyond the TMAD.

• If the TMAD is breached, the entire loss is covered by the SNRF. Funding Drawdown

• If a drawdown occurs, the SNRF will cover the loss.

• The agreement may be terminated if the strategy fails. Skin in the Game

• Eon Labs has skin in the game through TMAD

• Their skin in the game will start to decrease after the 11th cycle. External Researchers

• Currently partnering with one researcher named Sam

• Using a linear tree machine learning technique Inherent Capacity and Compounding

• Strategies have inherent capacity limitations

• Compounding may not be feasible beyond a certain point High Sharp Signals

• High sharp signals have lower realized vol and returns

• Capacity may be maxed out for high sharp signals Proposal for Absolute Dollar Value

• Consider changing percentage into absolute dollar value

• Absolute dollar value may make more sense in certain cases Payout and Infrastructure Costs

• Infrastructure cost are provided by Eon Labs Ltd. Absolute Dollar Numbers

• Absolute dollar numbers may not be favorable in some cases

• Returns need to be high for the framework to work well Signal Capacity

• The signal capacity is around 3 to 4 million.

• It is expected to hit the capacity within a month or two. Adjustments for Absolute Value Return

• The numbers will be adjusted to accommodate absolute value return.

• Suggestions for adjustments will be provided. Agreement Feedback

• No major issues were found in the agreement.

• Feedback on the language part will be provided later. High Capacity Signal

• There is another signal with a capacity of almost 100 million.

• It has a high potential return of 40% unlevered. Separate Discussion for high capacity strategy

• The discussion about the sharp strategy will be kept separate.

• Analysis and additional matrices will be shared. Quick Iteration for Going Live

• The goal is to go live with the signal as soon as possible.

• Agreement and trading can start after finalizing the iteration.

---

2023-06-28 3rd Interview Meeting

Video

A conversation about TradeFi expansion plans and research on different instruments, with a focus on crypto trading and commodity futures.

00:00 Discussing TradeFi’s expansion plans and Chen’s experience in trading commodity futures.

05:51 Trading strategy involves long and short positions to maintain dollar exposure neutral

10:44 Improving signal efficiency through better infrastructure

16:35 Experienced trader with background in electrical and financial engineering, spot FX, and futures trading now focusing on crypto signals.

21:50 External engagement for crypto signal, potential for future commodity futures work

26:47 Discussion on compensation and building infrastructure for commodity futures

31:35 We provide infrastructure for researchers to easily access data and run their systems.

36:28 Incorporating level one and level two depth snapshot features into existing model

---

Detailed Summary for Samrat Chen Terry (3rd session tech infrastructure) 2023 06 28 14 32 33 by Merlin

00:00 Discussing TradeFi’s expansion plans and Chen’s experience in trading commodity futures.

• TradeFi is a FinTech company that creates software for quantitative trading institutions.

• Chen has experience in trading various instruments, including commodities, and has used quantitative trading and machine learning in his trading strategies. 05:51 Trading strategy involves long and short positions to maintain dollar exposure neutral

• Rebalancing occurs every five minutes

• Strategy involves betting on spread of long and short coins 10:44 Improving signal efficiency through better infrastructure

• Desire for access to better data and execution infrastructure

• Previous experience building full pipeline and executing strategies 16:35 Experienced trader with background in electrical and financial engineering, spot FX, and futures trading now focusing on crypto signals.

• Started in FX spot trading desk at a bank, then moved to DRW and Exodus Point Capital Management trading spot FX and futures on global exchanges.

• Transitioned to crypto trading for independence and autonomy, managing small prop money and seeking partnership. 21:50 External engagement for crypto signal, potential for future commodity futures work

• High capacity signals have lower sharp

• Hybrid engagement proposed with Terry for internal work and external crypto signal 26:47 Discussion on compensation and building infrastructure for commodity futures

• Compensation structure will depend on whether the role is internal or external

• Infrastructure building includes handling symbology and choosing data provider 31:35 We provide infrastructure for researchers to easily access data and run their systems.

• Historical data is available for Bitcoin and Ethereum, with real-time feeds for other pairs.

• Access to data can be normalized for external researchers to reduce compute load.

• Real-time data is available for one-minute candlesticks. 36:28 Incorporating level one and level two depth snapshot features into existing model

• Current model is a statistical model with a machine learning layer for prediction

• Compensation details to be discussed later Hi Chen，

分享一位 candidate（身在紐約美國公民）信息

• Samrat Batth resume

——————

Samrat Batth, Ph.D.（Cerebro Search 獵頭 Michael 介紹）

Samrat，Terry 已經 meet 過兩次。

暫時收到各方提交 PnL 的成績中，Samrat 的績效排名第2-3 位。 有機會談成僱傭關係的 candidate 中，Samratt PnL 的績效第一。

其他前兩名的，一個是已有成熟融資體系的 Sakana（註：May 幫弄了一個 Sakana 對比 Eon 的績效參考）。

另外一個自稱 SMA 的三人團隊（Eon 打包僱傭三人的幾率可能不高）。

Samrat、Sakana 和 SMA 的績效差別不太大。

Sakana 和 SMA 有較長的 live trading 交易歷史。

Samrat 的 Live Trading 比較短，而且還原度不算高，因為 backtest 模型假設的是五分鐘調倉一次，而他 live trading 是一小時調倉一次。

扼要：

• Samrat 兩次提到閱讀到 Eon 在 Notion 發佈的學習資料，發現是他本人這些年來都有關注的，所以認為與 Eon 可能在探索 Quant 方面的認知同頻、可能能比較容易溝通
• Samrat 希望和 Eon 技術相關人員進一步 meeting 了解我們已有的基礎建設，試圖尋找能 leverage 他經驗和模型的空間
  • Terry 告知了 Samrat 關於 VIP9 的費率
• Terry 告知了 Samrat 關於 Eon 未來打算分工協作的規劃，但因為暫時還在規劃，所以暫時未能將 Samrat 對號入座。
• 因為信任和磨合都可能需要時間，為了清晰明確，可能先以外部 MSA 形式合作（類似 Sam 那樣）
• Samrat 會研究 Eon 的 profit sharing 框架和模板，之後再協商參數細節。 ————————

Chen，想請問你、May 或甚至 James 是否有空跟他碰一次？或安排誰先和他分享一下 infrastructure 相關資料呢？

Video

Video