• ulysseha19

Self-Tracking: Experimenting How to Experiment_V1.0

Updated: Jul 27

What is self-tracking in one sentence: self-tracking is a process and a tool to produce self-knowledge and optimise aspects of everyday life, by benchmarking and evaluating your habits and activities.

The objective of my mini-experiment in one sentence: use data collected from off-the-shelf consumer trackers (Oura and Dexcom G6) to understand what I want and don't want in my general routine.

The objective of this blog in one sentence: document and hopefully promote how a process/ methodology of quantifying daily activities can and should be applied to one's regimen, as opposed to passively digesting self-help tips and being left confused about whether something works.

0) Table of Content

Section 1 introduces the sociological problem I attempt to address by tracking myself, how, and why.

Section 2 describes my process.

Section 3 gives you the findings produced by the process.

Section 4 reflects on my process in real life (spoiler: failures and learnings)

Section 5 contains notes on how to improve my next experiment.

1) OH NO! My Stream of Consciousness is Fragmented! What to Do?

Changing ways of working

Only two screens

It is said that Carl Jung the psychiatrist built a retreat in St. Gallen in 1922 where he regularly visited for undisrupted, long sessions of deep work. Rising at 7 am and going to bed by 10 pm, a by-product of this rigorous routine was an "intense feeling of repose and renewal", according to Jung.

Incidentally, the topic of knowledge production, or labour, or intellectual "high" seems to have taken the opposite direction in our time, both in terms of task content (meetings/multitasking etc.) and the working environment (a hybrid reality of virtual and physical interaction). Hybrid 1 - Linkedin's new flagship office is one such example. With over 70 seating types to accommodate activities of different duration that a 'knowledge worker' engages in daily, ranging from a quick chat with colleagues in a semi-enclosed cubicle to more hard-hitting multi-screen setups for full-day immersive work.

Granted, the experience of intense euphoria derived from 'deep work' is a luxury that few enjoy. Yet, there might be a thing or two that can be improved in one's everyday professional life. A recurrent and most probably relatable experience, as a result of the changing knowledge production dynamics, is the cognitive burden of task-switching, which accelerates the erosion of sharpness and focus throughout the day. Switching attention between screens and objects in the screens over 2.5 to-do lists and 4 ad-hoc documentation media (Notion/notepad/actual notepad/ cell O24 in your spreadsheet etc.) guarantees out-of-body experience (spacing out) by lunch. Feeling tired all the time is not enjoyable.

Q: What to do about the sociological problem of lethargy?

A: Consider physiological intervention for a quick(ish) win.

A dude just tryna feel good

Over the past 2-3 years, I have experimented with different "dietary levers", including dietary practices (time-restricted eating like 18:6, biweekly 36h fasts, and slow carbs diet etc.) and supplementation ("nootropics" like noopept and off the shelf stuff like magnesium) to battle my long-lasting lethargy.

While the benefits of controlling certain variables are subjectively noticeable, especially (apparently) restricting my eating window, a systemic review of these variables is overdue, and a coherent routine or protocol is yet to be formed.

As such, I will use this article to examine and re-organize the existing levers at my disposal, based on their effects on my sleep and daytime energy fluctuation, using Oura and Dexcom G6 as my trackers. Ultimately, the objective is not to develop the perfect routine, but to start iterating, using this post as the pilot, both the levers as well as the methodology to examine their effect(iveness). In other words, it's about figuring out what to add to my routine, finding a way to measure and evaluate their effect, and keep working on the routine as long as it suits me.


If anything is to be done at all, it should be done with a reason. To track oneself is to learn about oneself, and I intend to use my self-knowledge to prolong and sustain my subjective feeling of sharpness. While I find the general fitness and productivity culture overall aesthetically unpleasant, it is undeniable that many benefits can be reaped from an industrious lifestyle that is maximally insulated from ailments. It is true that likening lethargy, an underlying discontent, and the way that we work might be a stretch. Nonetheless, whatever makes you feel slightly better should be considered a win. Fundamentally, I am doing this out of boredom, with the hope that a side project would serve as an intellectual anchor in my post-university life.

2) My Method of Self-Tracking


The image above demonstrates my overall approach. To me, a good day is the opposite of lethargy. I define feel good as the subjective feeling of sharpness throughout a day (which necessarily entails self-reported data), which is considered to be a result of decent sleep quality and a decent diet, (with 3 'levers' to consider: what, when, and how much) according to the prevailing wellness truism.

Ultimately, my goal is to hopefully identify meaningful correlation between what I eat plus how I sleep and how my feeling fluctuates throughout the day. Consumption is further broken down into supplementary and dietary, each with two specific variables that are measured by different metrics. Whereas supplementary variables include actual supplementation (e.g., magnesium) as well as caffeine consumption (the 3 levers also apply), dietary variables refer to eating window (fasting) and carbs ("fast" vs. "slow" carbs and sugar).

My hypothesis is that blood sugar level spikes would slow me down (which proves to be inconclusive). Therefore, I should keep it consistent and avoid any food that spikes my glucose. Meanwhile, caffeine is known to have a 5-7 hours half-life and its detrimental effect on sleep is well-documented. By examining my coffee intake, I hope to find an acceptable substitute and ideal cut-off time.


Correspondingly, different tools will be used to measure my body's reaction to things that I put in my mouth as well as my sleep. For instance, Dexcom G6, a continuous glucose monitoring (CGM) device, will be used to measure my blood glucose level against my feeling (self-report) of alertness and clarity after a meal. Oura, on the other hand, will be used to measure my heart rate, a proxy of my reaction to caffeine, and my sleep quality, on which supplementation and caffeine control should have an effect.


A baseline will first be established with a constraint-free diet in the first week (e.g., no time restriction, no caffeine cut-off time, much carbs as I like), with more dietary principles to be added subsequently (e.g., time-restricted eating plus time-restricted caffeine consumption, restricted amount and types of carbs). Each block is roughly 10 days, which is the life cycle of a Dexcom G6 receptor, depending on the time of activation.

3) General Key Findings

#1 Cardio seems to be yuuge in regulating glucose

Average glucose (mmol/L) by block: 6.2/5.9/6.9

My glucose level averaged the lowest (5.9) in the second block as I incorporated some jogs in block 2, despite my lenient dietary approach (white grains/sugar). To roughly benchmark the 'goal range' of blood glucose level with reference to Dexcom's material, I should aim for <7.8 mmol/L.

Using one extreme example from block 2, after a not particularly 'clean' lunch (lots of processed white carbs), my blood glucose level peaked at 8.2 mmol/L approximately one hour after the meal, going from a 4.5-5 mmol/L neighborhood. As I started jogging, it dropped to as low as 4.3 mmol/L and stayed within the range of high 4/low 5.

By contrast, a glucose spike was observed on a non-cardio day, peaking at 11 mmol/L roughly 1.5hr after dinner.

#2 But the correlation between glucose and perceived energy level remains unclear

The correlation between glucose and perceived energy level remains unclear. I often felt clear-headed despite post-dinner glucose spikes. Meanwhile, low glucose (5-6mmol/L) in a morning fasted state, using the below weekly overlay as an example, could co-exist with drowsiness.

The effect of fasting on productivity seems to be chiefly psychological and the emancipation of time, in that impact on morning-early afternoon sharpness was almost unnoticeable.

#3 Swap coffee with green tea for a more sustainable 'high'

In terms of caffeine choice, swapping coffee with tea is also yuuge for a more sustainable subjective ‘high’ throughout the day.

#4 Time-restricted eating has more of a psychological impact on productivity

Fasting (16:8) had more of a psychological effect on productivity in that I didn’t think about food while fasting. It also gives me more time, especially during longer fasts (36hours etc.).

4) Reflection on Execution

Looking back, this past mini experiment has, unexpectedly, produced more insights into my methodology than identifying a set of items to optimise energy fluctuation and/or sleep. Specifically, I have failed to account for externalities such as social interaction and my circle's penchant for festivities. There was no way to ensure comparable sleep data, for instance, after a heavy night out that was triggered by 2 beers. As such, my biggest takeaway is as much about realising the critical significance of methodological consistency (next steps in section 5) as discovering the value of cardio and caffeine substitutes. I outline below my reflection by block.

Block #1 - Jan 25th - Feb 4th

Key externality: Chinese New Year

All in all, a block that was filled with social activities and heavy drinking (festivities during Chinese New Year to be blamed).

A key learning of block 1 is the crucial role of methodological consistency as findings in the following aspects were undermined:

  • Sleep (sleep deprivation and irregular hours)

  • Glucose (over and underfeeding myself)

  • Subjective feeling of sharpness (compounded hangovers).

Block #2 - Feb 21st - March 3rd

Key externality: Lockdown

I had the opportunity and better control my diet during lockdown in HK but still failed to adhere to a slow carb/minimal trash intake diet.

The key methodological learning was to keep a food journal (or phone in the absence of a journal) during the day and map the entries (food items consumed) onto my glucose chart.

Lesson learned:

  • My glucose level was consistently kept at a lower range (7-9 mmol/L) throughout this block, but I didn't feel particularly sharp.

  • I begin to question the extent to which 'aliveness’ and ‘sharpness’ can be optimized, in the absence of intellectual, ideological and/or psychological stimulation.

  • Meanwhile, the idea of sleep began to feel stressful with the expectation of positive effects from fasting and supplementation.

Block #3 - April 7th - April 17th

Key externality: Pre-open up excitement

The worst of the 3 blocks in terms of consistency, as I got sucked into a hedonistic quagmire against a summerly, almost-out-of-lockdown backdrop, which culminated into 4 big nights over 10 days.

No data was captured, with the dietary protocol out the window. Total chaos

Lesson learned:

  • Experimental commitment is absolutely essential. If you are going to test out ahypothesis in a given period of time, you have to adjust your lifestyle.

  • Lifestyle optimization necessarily means standardizing the components and activities of life in a somewhat sterile manner. Therefore, my general routine applies mostly during workdays, in the absence of social activities which are typically at odds with sleep quality/energy fluctuation control. Reforming the nature of my social life is a decision I’m admittedly not yet ready to make.

5) For Next Iteration...

This series constitutes the first of my attempts to develop a general, workday routine. This means the learnings from here will be taken to improve the subsequent iterations of my routine.

#1 Standardisation

Basically a version of A/B testing to enable data consistency and comparison.

Eating and caffeine consumption window must be the same to produce more directly comparable data. For instance, glucose data that could be fit consistently in the same territory of an overlay chart, and the impact of caffeine on sleep from tea vs. coffee (provided that bedtime and duration are also consistent).

#2 Make it easier, establish instructions and follow them

The biggest failure of this mini project is of a methodological nature. Perhaps the most effective solution is to plan out detailed daily instructions à la Biden to mechanically follow next time. For instance, "eat lentil stew at 1300, finish meal by 1320, and take 100mg of magnesium" or "screens off by 2100, read for 45 minutes then try to fall asleep".

#3 Central planning

The plan should be centrally planned (e.g., every Sunday for the following week/ 2 weeks) and include weekend social activities or just exclude them altogether.

6) Appendix

6.1) Sleep

Average sleep score: 75/76/70

Average sleep efficiency: 77%/75%/67%

Average HRV: 39/48/46

Average coffee: 0.5/1.2/0.9

6.2) Energy Fluctuation


Halfway through block 2, I decided to stop tracking heart rate. A quick google search will tell you that a single espresso shot contains 70-80 mg of caffeine, whereas a bag of green tea from Lipton contains 28-38mg. It doesn't take Sherlock to guess that coffee impacts my heart rate more than tea. Tea typically induces a smaller heart rate spike (60-90 BPM compared within 3 hours of consumption, compared to coffee which could lead up to >100 BPM).

All in all, subjectively speaking, the high from tea is generally more sustainable.


Ambiguity is found not only in the correlation between blood glucose level and sharpness, but also the impact of food on blood glucose spikes. For instance, glucose spikes have been caused by supposedly ‘cleaner’ food (e.g., sweet potato, brown rice with chicken), while, as previously seen, I could get away with eating 'bad' carbs such as pasta and large amount of sugar.

29 views0 comments