Kavod Holdings
Quantitative Investment Management
Kavod Holdings is an Emerging Short-Only Investment Management Firm, which utilizes advanced statistical analysis, applied mathematics, artificial intelligence, and market expertise to identify profitable investment opportunities. The partners have been collaborating since 2019, dedicating their efforts to developing and refining their algorithms and machine learning models.
System Overview
Short Only Algorithm
Trading decisions are driven by a precise and sophisticated Trading System (The System). The System is composed of multiple individual segments, each one designed to serve a unique purpose. Though single minded by design, each individual segment is intricately interconnected, synergistically working together to derive alpha from our signals.
Trading Methodology
Investment Universe
● Equity Stocks on NYSE
& Nasdaq
● $15 - $400 per shares
● 2.1 MM daily volume
● >19 days historical
Ranklist Generator
● Monitoring Assets Through Trend Cycles to isolate the beginning of Asset Drawdown
● Amount of assets on each Ranklist Fluctuates Given Market Cycle
AI Driven Decisions
Two Advanced Artificial
Intelligence Models govern the following processes:
● Risk Management
● Trend Selection
AI Trend Selection
Dataset Information

● Dataset Date Range: 1/2/07-5/12/22
● 400 Assets = 209,577 Trends
● 5 Downtrend Groups (Downtrends generated By Short Only Algorithm):
a. Poor Negative: >0%
b. Poor Positive: 0-1.5%
c. Regular: 1.5-5%
d. Regular Long: 1.5-5% and < (Length Constraint)
e. Good >5% and > Length Constraint
f. Ideal: >5% and < Length Constraint
SR Composition
SR represents the mathematical quantification of an event that occurs prior to significant downtrends. For our System to initiate a short position, it must predict the presence of a genuine SR within a specific proximity. Notably, the strength of the SR signal itself is empirically supported by data, demonstrating its ability to generate alpha.
SR Profitability
To gauge the profitability of the SR quantification, we executed an experiment with the following constraints.
● 2012−2018 Trades selected by Ranklist Generator and Meta Model
● Trade Liquidation is decided in hindsight with a constraint that the downtrend cannot lose 0.4% of its profits calculated from the last close.

For more information on this experiment, please request the white paper "Strength of Signal".
● Date range: 2012-2018
● Traded: 4,467
● Greatest Individual Profit: 60%
● Greatest Individual Loss: 2.5%
● Annualized Return: 109%
● Sharpe: 10.81
AI Performance Metrics
The MetaModel
Within The System exists a group of state-of-the-art deep neural networks known collectively as the "Meta Model". The Meta Model is designed to select Assets most likely to contain a real SR (i.e true positive) within a certain time frame with a high degree of Precision (74%) and Accuracy (80%) through 15 years.
Portfolio Management Systems
Risk Management Methods
First Positive — In case the Meta Model has a prediction error (False Positive), we have an Algorithm called "The First Positive" which handles these potential trading losses. The first step of this model is a logistic regression process.

75% of remaining Poor-Negative (i.e losing trades) Downtrends are eliminated after 1 time interval (15 minutes), thus limiting the loss from losing downtrends to an absolute minimum.
Backtest Performance
Select range test from 2012–2018
This Backtest was chosen to depict likelihood of loss in extended periods of bullish activity.

● Date range: 2012–2018
● Traded: 447
● Total Return: 131.69%
● Benchmark Return: 85.16%
● Greatest Individual Profit: 23%
● Greatest Individual Loss: 7.0%
● Sharpe Ratio: 5.40
● Average Winning Trade: 4.19%
● Average Losing Trade: -0.54%
Michael Friedman
Gabriel Kingsley-Nyinah
General Disclaimer
This communication may contain an investment recommendation, however, Kavod and all members disclaim all liability for providing this investment recommendation and accept no liability whatsoever for any direct, indirect or consequential loss arising from its use. Not all investment strategies are appropriate at all times, and past performance is not necessarily a guide to future performance. Any opinions or estimates expressed herein reflect the judgment of the author(s) as of the date the investment recommendation was prepared and are subject to change at anytime without notice. Unless otherwise stated, the information or opinions presented, or the research or analysis upon which they are based, are valid at the point of publication, as they are not updated in real time to take account of changes in the financial markets or new news about the issuer or financial instruments.