diff --git a/docs/observables.md b/docs/observables.md
index a35ccf0..eafcd1d 100644
--- a/docs/observables.md
+++ b/docs/observables.md
@@ -1,15 +1,66 @@
+<div align=center>
+  
 # What does hdspin calculate?
 
-We outline the various observables that hdspin calculates.
+</div>
 
-- Statistics are not saved at every timestep. As it is possible to have many millions of timesteps, this is not tractable from a storage standpoint. Grids are generally linear in log-space, and observables are saved only at those timesteps.
-- Most observables have a variety of statistics saved, including the mean, standard deviation, standard error, and median (at every timestep on the grid).  
-- All averaged statistics can be found in the `results.json` file saved at the end of the simulation.
+**We outline the various observables that hdspin calculates.**
 
-## Average energy
+> [!IMPORTANT]
+> Statistics are not saved at every timestep. As it is possible to have a _huge_ number of timesteps, possibly on the order of many billions, this is not tractable from a storage standpoint. Grids are generally linear in log-space, and observables are saved only at those timesteps. Most observables have a variety of statistics saved, including the mean, standard deviation, standard error, and median (at every timestep on the grid).
+
+All averaged statistics can be found in the `results.json` file saved at the end of the simulation. You can load the results via the built-in `json` library in Python via something like this:
+```python3
+import json
+with open("results.json", "r") as f:
+  results = json.load(f)
+```
+Similarly, diagnostics are saved in `diagnostics.json` and the original config is saved in `config.json`. These files can be loaded simiarly. You can access the available observables via `results.keys()`, and for a given observable, such as the energy, you can access the available statistical information via `results["energy"].keys()`.
+
+# Energy
 
 Consider the energy of tracer $i$ at simulation clock time $t$ is $E_{i}(t).$ The average energy over $N$ tracers is simply given by
 
 $$ E(t) = \frac{1}{N} \sum_{i=1}^N E_i(t).$$
 
-Results for the energy can be accessed via `results["energy"]`.
+Results for the energy can be accessed via `results["energy"]`. Similarly, the grid on which the energy is recorded can be accessed via `results["grids"]["energy"]`. The standard deviation, and standard error, are similarly calculated and accessed,
+
+$$ \sigma_E(t) = \sqrt{\frac{1}{N} \sum_{i=1}^N (E(t) - E_i(t))^2}.$$
+
+All in summary, one can access these variables via:
+
+```python
+key = "energy"
+grid = results["grids"][key]
+energy = results[key]["mean"]
+energy_standard_error = results[key]["standard_error"]
+```
+
+# E-max
+
+TK
+
+# Ridge energy
+
+# Autocorrelation functions ($\psi$)
+
+TK
+
+# Aging functions ($\Pi$)
+
+TK
+
+# Walltime-per-waiting time
+
+TK
+
+# Acceptance rate
+
+The cumulative acceptance rate is logged on the same grid as the energy. It represents the Metropolis acceptance rate up until that point in time, and can be accessed via `key = "acceptance_rate"`. Defining the cumulative acceptance rate at time $t$ for tracer $i$ as $A_i(t),$ it is given by the ratio of the number of accepted steps to the number of total steps. Note that in Gillespie simulations, $A_i(t) = 1.$ The statistical moments of the cumulative acceptance rate are calculated in the same fashion as the energy.
+
+# Cache size
+
+TK
+
+
+